[NEW MERCENARIES] Please note that all new forum users have to be approved before posting. This process can take up to 24 hours, and we appreciate your patience.
Because AS WAS LISTED IN THE LINK IN MY PREVIOUS POST Meltdown has only 1 KNOWN variant. The claim that AMD is IMMUNE to Meltdown is false because it is premature; AMD is highly likely to be susceptible to ANY variant of Meltdown that is currently unknown. And since these companies were so stupid as to make the exploits public, any random joe can find 5 million freaking variants to this thing, including multiple Meltdown variants.
You need to ****ing read. I've already said this before, but you don't seem to understand - known variants =/= all variants. One variant of Meltdown is not indicative of all variants of Meltdown. AMD used what little good news they could to trick the public into believing the lie that it's Meltdown immune; the ONLY tests done on it were for one variant that the testers themselves said was incomplete, as they had yet to do more precise testing.
Hopefully you'll actually read the article this time. But if not, just ask literally anyone else that's talked about the cores before on these forums: More does not mean better. If a program is not designed to make use of multiple threads, then it won't. Most programs don't even efficiently use quad core.
Literally the ONLY reason you'd get a quad, octo, whatever core is if each thread has the capacity to handle any single-threaded program in a single thread without problems. That way you can just separate tons of programs into tons of threads without worry, and everything's better organized. But if you think a processor with thirty-two threads can handle that, you are really puffing the dragon.
What are you talking about? All of the companies have said they're making patches. Intel and Nvidia said they'll have patches within the week. They even specifically stated that the patches won't fix the issue, multiple times even.
Those articles stated that every single CPU is affected by Spectre, and anyone with common sense would know that Meltdown can have new variants that affect all CPUs as well. Not just Intel's; everybody's.
All of the companies have also said they're re-designing future chips to not have the issue. Future chips, as in not any chips that are currently set to launch. Especially since they don't know how they'll redesign the chips yet to not have the exploits.
It seems that you need to go back and read the articles, too, because that information is provided in almost every single one of them.
I seriously don't even know why I'm wasting my time with people who aren't willing to do the research. Go ahead and buy the newest super fancy Ryzen and deal with the dozens or hundreds of variants of the exploits, if you're so certain it'll be immune.
Not buying a CPU for gaming because there is a design flaw in it that actually speeds up things is just i don't know what to call it. This is basically a flaw of sacrificing security for speed.
I'll sum it up for programmers to understand the main part (side-channel part) of the Spectre/Meltdown somewhat more easily (it took me a few hours to understand):
This part of the exploit code uses two arrays, one for the out-of-bounds access (small array with like 16 elements) and the other one is used for testing the indexes after (large array with like 256 * 512 elements). It needs a large array because the value of the protected memory address will be copied into it as an index by the speculative execution behaviour of the CPU that will discard this memory content from the registers after it refuses giving access to it for the application. The large array however still holds a memory address that is unreachable for the application but it doesn't need it because the memory content is "stored" as an index of the large array. I quoted it because it's not really stored, it just still holds the protected memory address in the cache (thus faster read time). By probing (dereferencing) each index of the array and measuring the time of each probe you can retrieve the byte it holds from the array index. (so the key part is the large array's index)
Example code from exploit-db.com:
// array1 is the small array
// array2 is the large array that holds the value
// x is the out-of-bounds offset
if (x < array1_size) {
temp &= array2[array1[x] * 512];
}
This is a real threat however there are no malwares specialized for this so no need to worry yet. And it's quite difficult to fix completely or without heavy performance losses.
Meltdown has one variant. Spectre has two variants.
1 + 2 = 3
The claim that AMD is IMMUNE to Meltdown is false because it is premature;
Out of the 5+ tech forums I browse, you're actually the only one to think this. You accuse others of advertising AMD, but now I find it difficult to believe you're not being paid by Intel. If you think there are more than three variants, please do list them. Once again maybe there are six variants or even more but we are only aware of three so we'll just have to accept it and move on. All shall be revealed soon enough anyway.
Assuming the same architecture and clock speed, more cores are better. 6 Coffeelake cores > 4 Coffeelake cores. Really not that hard. That article is from 2015 and for mobile devices, we're in 2018 and on x86 hardware so it's completely irrelevant. Some people use their computer for more than games, stuff that can scale to 32 threads and even more.
We are indeed in the multithreaded era, here is an example of Crysis 3, a 2013 game.
The quad core i5 still delivers a decent experience, but it doesn't change the fact that practically every AAA game you buy today uses at least 6 threads. I know from personal experience every Assassin's Creed game since ACU uses at least 8 threads.
Not buying a CPU for gaming because there is a design flaw in it that actually speeds up things is just i don't know what to call it. This is basically a flaw of sacrificing security for speed.
This is a real threat however there are no malwares specialized for this so no need to worry yet. And it's quite difficult to fix completely or without heavy performance losses.
Well, one company said they felt Meltdown - or rather, the specific part of the problem that allows for the Meltdown exploit - slowed down their game's performance, as the "patch" they released specifically for their game showed better performance afterward, though exactly how much is debatable.
Don't ask me how. I haven't seen their code, and I likely wouldn't understand it.
If you think there are more than three variants, please do list them. Once again maybe there are six variants or even more but we are only aware of three so we'll just have to accept it and move on. All shall be revealed soon enough anyway.
"soon"
No. That's how this works.
I stated, again and again omg, that there are three KNOOOOOOOOOOOOOOOOOOOOOWWWWWWWWWWWWN variants. KNOOOOOOOOOOOOOOOOOOOOOWWWWWWWWWN. How hard is that to understand? It's a simple ****ing word.
But sure, accept it and move on while AMD fails to mention the 20+ new variants in the next year that will allow people to spy on the chip that it won't fix.
Assuming the same architecture and clock speed, more cores are better. 6 Coffeelake cores > 4 Coffeelake cores. Really not that hard. That article is from 2015 and for mobile devices, we're in 2018 and on x86 hardware so it's completely irrelevant. Some people use their computer for more than games, stuff that can scale to 32 threads and even more.
The quad core i5 still delivers a decent experience, but it doesn't change the fact that practically every AAA game you buy today uses at least 6 threads. I know from personal experience every Assassin's Creed game since ACU uses at least 8 threads.
Crysis and Assassin's Creed are benchmark games. Even among AAA games, they're benchmarks. And don't give me that "at least 8 threads" - no. It uses at least ONE thread, and can potentially multi-thread beyond eight. You don't seem to understand how hyperthreading works.
Let's say there's a Single Core, and it's equal to 1000. An equivalent Dual Core, is 500+500. An equivalent Quad Core is 250+250+250+250. What we really care about is threads, though, and we determine one core as equal to two threads.
Now, hyperthreading is super awesome, but it doesn't care how many threads there are, so long as there's more than one. There is no performance difference overall regardless of whether the threads are 500+500 or 125+125+125+125+125+125+125+125. There's more threads to put stuff onto, sure, but all that matters is that we can use hyperthreading properly at all.
With this in mind, understand that Quad Core is already a gimmick. Single Core meets all of our needs, and will met all of our needs, regardless of whether we're playing Stardew Valley or Crysis 20,000. The only exception is games like Vindictus, which are meant to run on Single Core because they take up so much.
Then again, Vindictus can run just fine in a single thread out of my eight threads, because each thread can handle that much. Vindictus can run on 125. That's great.
But you seem to think that not only l33t new AAA games, but even old games like Vindictus will work fine on a thirty-two thread chip. This means you think, out of the sum total of 1000, and the equation of 31.25+31.25+31.25+31.25+31.25+31.25+31.25+31.25+31.25+31.25+31.25+31.25+31.25+31.25+31.25+31.25+31.25+31.25+31.25+31.25+31.25+31.25+31.25+31.25+31.25+31.25+31.25+31.25+31.25+31.25+31.25+31.25, that Vindictus can run on just 31.25.
No. And that you think your 8+ AC game will be able to smoothly play on an equivalent 32 is even more ridiculous.
While I highly doubt Ryzen will let any of its threads only handle that small of an amount, it doesn't change the fact that even IF a single thread on Ryzen could handle 200, its competitors with single core - if they chose to have two threads on that core to allow for hyperthreading - would have 3200+3200 aka 3200 per thread which could handle any number of programs regardless of whether those programs can hyperthread or not.
Increasing the number of threads does not magically increase performance. Hyperthreading itself - just hyperthreading - increases performance.
Wow, you learned something new today!
Oh, and a core can have more than two threads on it. Mind=blown, right? So Quad Cores could totally have thirty-two threads...but they don't, because it's a terrible gimmick.
Rezi doesn't seem to deal with facts... Even though you made a very easy to understand post I wouldn't waste your time with someone that doesn't want to hear the facts if it doesn't side with their opinion. Let them believe what they want there is plenty of information out about it disputing Rezis misguided information.
Oh and last time I checked amd has had processors out for a very long time and they don't suffer from the meltdown bug.. so what's with the premature stuff??
I wouldn't bother replying to this thread anymore. Any intelligent being reading this post from the beginning will understand the message. Going back and reading all the comments, i realized Rezi replied "
AMD, Google, Intel, all of them have the same thing. It's not like AMD magically made their CPUs load fast without needing to pre-load.
Also, sounds like they believe Google makes CPUs.
And after reading their recent posts, it seems they don't have any kind of degree related to computer science nor is knowledgable in the subject area.
They do not seem to understand why AMD is fundamentally IMMUNE to meltdown exploit. The way AMD operates is different and AMD is IMMUNE to all variants of meltdown. Perhaps another different unknown exploit exists for AMD CPUs, however worrying about an exploit yet to be discovered is like worrying about World War 3. It's no use trying to explain to them; they won't understand/accept it.
@LeXicO Thank you for your concern but this is more for humanity's sake than an individual's.
I stated, again and again omg, that there are three KNOOOOOOOOOOOOOOOOOOOOOWWWWWWWWWWWWN variants. KNOOOOOOOOOOOOOOOOOOOOOWWWWWWWWWN. How hard is that to understand? It's a simple ****ing word.
But sure, accept it and move on while AMD fails to mention the 20+ new variants in the next year that will allow people to spy on the chip that it won't fix.
Yes there are three known variants, every tech forum I visit agrees with this. However, you seem to think there are more than three and if you are really think this is the case, please tell us because I'm sure everyone will be very interested. But as far as I can tell, you're making this up due to the fact Intel is more screwed than AMD at the moment.
Crysis and Assassin's Creed are benchmark games. Even among AAA games, they're benchmarks.
They get benchmarked because they're demanding. They get benchmarked because people might need to upgrade their PC hard to play it. They get benchmarked because they show the difference in performance of PC hardware. Crysis 3 is chosen by DigitalFoundry because it scales very well with extra threads. They get benchmarked because people actually play these games. Maybe not Crysis 3, but games such as Witcher 3, GTA 5, BF1 and ACO are popular, scale well and demanding. What do you suggest, benchmark Minesweeper?
And don't give me that "at least 8 threads" - no. It uses at least ONE thread, and can potentially multi-thread beyond eight. You don't seem to understand how hyperthreading works.
Of course it uses at least one thread, what doesn't? Yes, ACU uses at least eight threads.
Let's say there's a Single Core, and it's equal to 1000. An equivalent Dual Core, is 500+500. An equivalent Quad Core is 250+250+250+250. What we really care about is threads, though, and we determine one core as equal to two threads.
Now, hyperthreading is super awesome, but it doesn't care how many threads there are, so long as there's more than one. There is no performance difference overall regardless of whether the threads are 500+500 or 125+125+125+125+125+125+125+125. There's more threads to put stuff onto, sure, but all that matters is that we can use hyperthreading properly at all.
With this in mind, understand that Quad Core is already a gimmick. Single Core meets all of our needs, and will met all of our needs, regardless of whether we're playing Stardew Valley or Crysis 20,000. The only exception is games like Vindictus, which are meant to run on Single Core because they take up so much.
Here is what happens when GTA 5 is ran on a single core CPU:
@3:17
The built-in GTA 5 benchmark didn't even finish and Ashes of the Benchmark averaged less than 1 FPS. I'd say a single core CPU does not meet all our needs.
But you seem to think that not only l33t new AAA games, but even old games like Vindictus will work fine on a thirty-two thread chip. This means you think, out of the sum total of 1000, and the equation of 31.25+31.25+31.25+31.25+31.25+31.25+31.25+31.25+31.25+31.25+31.25+31.25+31.25+31.25+31.25+31.25+31.25+31.25+31.25+31.25+31.25+31.25+31.25+31.25+31.25+31.25+31.25+31.25+31.25+31.25+31.25+31.25, that Vindictus can run on just 31.25.
No. And that you think your 8+ AC game will be able to smoothly play on an equivalent 32 is even more ridiculous.
Where did I say this? Regardless, I'm fairly certain Vindictus runs fine on a Threadriper 1950X. Probably not as a good as an i7 8700k, but it should be good enough. Also games that thread well such as ACU run quite nicely on a Threadripper 1950X.
If a Threadripper 1950X can push a GTX 1080 Ti to 95%+ usage, I'd say that's pretty good.
While I highly doubt Ryzen will let any of its threads only handle that small of an amount, it doesn't change the fact that even IF a single thread on Ryzen could handle 200, its competitors with single core - if they chose to have two threads on that core to allow for hyperthreading - would have 3200+3200 aka 3200 per thread which could handle any number of programs regardless of whether those programs can hyperthread or not.
Increasing the number of threads does not magically increase performance. Hyperthreading itself - just hyperthreading - increases performance.
It depends on the application, but Hyper-threading is definitely not some kind of magic that increase performance for all applications, especially single core applications.
Edit: Here are the results of what happens when I disable SMT of a Ryzen 7 1700 CPU.
As expected, single core performance is unaffected. I would have done it with an Intel CPU, but I don't have a Hyper-threaded Intel CPU available at the moment.
Oh, and a core can have more than two threads on it. Mind=blown, right? So Quad Cores could totally have thirty-two threads...but they don't, because it's a terrible gimmick.
Consumer CPUs don't have eight threads per core because consumer applications aren't designed for them. Not a gimmick, these CPUs exist for a reason.
Some say they couldn't do the exploit on AMD and some ARM processors because they didn't optimize the code enough, so basically it ran slower on them. Since there is a time limit for cache hits, because they can't be slower than cache misses, they can barely infer its contents from time measures. But this one is just speculation because i don't know how AMD chips work.
The problem is that speculative execution doesn't check for memory access violations to speed up the branch prediction. And because it's a speculation, it just executes the predicted code not being sure in any results, because the main execution is still waiting for memory reads because of a prior cache flush, so it can continue after the memory contents are loaded into the cache.
Processors would be much slower without this mechanism so i don't know how AMD solved this problem but reading from protected memory is probably a threat to AMD processors as well. The only processors invulnerable to these are the ones that don't utilize speculative execution, so they don't execute code predicted by the branch predictor unit, they only load the instructions. For example many Raspberry Pi series like the Raspberry Pi 3 Model B that uses an ARM Cortex-A53.
Hi Noobie, in answer to your post, it sounds as if your running similar hardware to my system. I use a gigabyte motherboard. I had this system custom built back in 2013 and I run an amd 8120 8 core processor. the game overall for me runs fine. so the processor and video card you are listing should work fine. just FYI the motherboard I have is a GA-970a-D3p ver 1. I hope this helps.
I still have to tune down some of the ingame video setting to maintain 60fps despite having a monster overclock. Usally fps dips in raids or crowded city although cpu nor gpu being used that much. Mostly between 20% - 70% across all cores. Same game with the gpu. The optimization is just bad.
If you wanna go for an amd cpu, pick Ryzen 7 1800x and overclock the hell out of it. A mediocre gpu is enough for Vindictus. Just my 2 cents.
I have the FX 8350 paired with a Gigabyte GA-970A-DS3P FX motherboard. You should be fine upgrading from the FX 4300. I also have 32GB of ram and XFX RX 560 for graphics.
AMD, Google, Intel, all of them have the same thing. It's not like AMD magically made their CPUs load fast without needing to pre-load.
Also, sounds like they believe Google makes CPUs.
They do not seem to understand why AMD is fundamentally IMMUNE to meltdown exploit. The way AMD operates is different and AMD is IMMUNE to all variants of meltdown.
The OP asked which AMD processor to get. Just processor. Then he stated he has a gtx already, which is a GPU. So I figured he was the one confusing CPU and GPU and was actually looking for a GPU.
All CPUs work the same. Your claim is as ridiculous as if one uses analog radio waves and the other uses binary code. All CPUs have company-unique architecture, but all of them function in the same manner that allows for both of these exploits. I've repeated this multiple times, and even linked articles that explained the situation perfectly.
Yes there are three known variants, every tech forum I visit agrees with this. However, you seem to think there are more than three and if you are really think this is the case, please tell us because I'm sure everyone will be very interested. But as far as I can tell, you're making this up due to the fact Intel is more screwed than AMD at the moment.
Okay, you apparently don't seem to understand what the word "known" means. You need to get a dictionary.
Known means found. Only three variants have been found, only one for meltdown.
But more CAN be found, the UNKNOWN variants. Immunity to a single variant does NOT represent immunity to ALL variants.
AMD is immune to a SINGLE variant. Yet it claims to be immune to the entire exploit. This is a lie, because someone CANNOT claim immunity to variants that are not YET known.
Logic.
They get benchmarked because people actually play these games. Maybe not Crysis 3, but games such as Witcher 3, GTA 5, BF1 and ACO are popular, scale well and demanding.
A benchmark game is a game that attempts to make full use of known hardware/software. It's not that hard of a concept.
My point was that you seem to think that benchmark (or by your terminology, "demanding") games represent the entire game industry, and possibly even other programs; they don't. Just because one game can hyperthread on eight threads, doesn't mean they all can; Vindictus certainly can't. What you seem to think is that Vindictus, or even games that can hyperthread on, say, four threads, can hyperthread on thirty-two threads. No; that's not how it works.
Of course it uses at least one thread, what doesn't? Yes, ACU uses at least eight threads.
Get a dictionary. Seriously. Look up "at least" - "at least" means "bare minimum" - if a game used "at least" eight threads, that would mean it either can't use MORE than...or can't use LESS than eight threads, depending on the context of the sentence. No; it can use "bare minimum" one thread, just like every other program.
I never argued against this. You didn't pay attention to what I said.
It is possible for some AAAAAAAA game to hyperthread through thirty-two threads. The whole point of multiple threads is, of course, to increase performance by decreasing stress a la Amdahl's Law.
But just because ONE game can do that, doesn't mean they ALL can. A game can hyperthread a certain number of threads depending on how it's made. A game that can hyperthread only two threads can't hyperthread eight; a game that can't hyperthread at all, like Vindictus, certainly can't hyperthread eight, and especially not thirty-two.
Having thirty-two threads is only good if any program can operate fully on a single thread of those thirty-two, because otherwise you're limiting yourself to programs that can operate on a large number of threads. By contrast, a core with two threads can run any game with or without hyperthreading just fine; there's less ability to use Amdahl's Law, but more space in each thread, which means you're not compromising the number of processes that can be run.
@3:17
The built-in GTA 5 benchmark didn't even finish and Ashes of the Benchmark averaged less than 1 FPS. I'd say a single core CPU does not meet all our needs.
You should have watched your own video. Single Core =/= single thread. The guy intentionally disabled hyperthreading, even though you can use hyperthreading with just two threads, which most single cores now have.
The video is wrong, because the test is done incorrectly. GTA V has trouble without hyperthreading, of course because hyperthreading increases performance. But the tester disabled hyperthreading despite that being completely unnecessary, which made the audience THINK that the performance hit was due to only having a single core, when in fact it was because the tester forced the game to run on a single thread.
It depends on the application, but Hyper-threading is definitely not some kind of magic that increase performance for all applications, especially single core applications.
This is MY argument, not yours. Literally this is my whole argument - that a thirty-two thread CPU won't run Vindictus well at all.
Consumer CPUs don't have eight threads per core because consumer applications aren't designed for them. Not a gimmick, these CPUs exist for a reason.
This is ALSO my argument. Consumer applications aren't designed for a CPU with thirty-two threads. Now you're just arguing against yourself.
At this point I just think you're trolling. You link things without reading/watching them. You are even trying to argue points against me...that I've already argued against you. You're failing to use basic logic, and ignoring every well-explained point I'm putting across.
All you're doing is misleading the OP and other users in your attempt to promote AMD, a company that has repeatedly lied, cheated, and cheapened over the years. And I've provided every single explanation necessary to show how much its hardware is a gimmick.
Hopefully the OP will be smart enough to check how much each thread can handle, instead of just going after the number of threads. If not, it's his loss.
Okay, before anyone ****posts in response to my previous post, it looks like you CAN hyperthread with Vindictus, it's just not designed to do so, so you have to force it. And I'm not exactly sure what it even gets out of hyperthreading. Anyone have data for that?
I just tested it through Steam, and it actually decreased performance for me, but it might be different for someone else.
All CPUs work the same. Your claim is as ridiculous as if one uses analog radio waves and the other uses binary code. All CPUs have company-unique architecture, but all of them function in the same manner that allows for both of these exploits. I've repeated this multiple times, and even linked articles that explained the situation perfectly.
If all CPUs work the same, please explain why AMD think they are so confident in their CPUs unlike Intel. I'm not a CPU engineer but if you think you have the knowledge to know why they are, feel free to explain in detail.
Okay, you apparently don't seem to understand what the word "known" means. You need to get a dictionary.
Known means found. Only three variants have been found, only one for meltdown.
But more CAN be found, the UNKNOWN variants. Immunity to a single variant does NOT represent immunity to ALL variants.
AMD is immune to a SINGLE variant. Yet it claims to be immune to the entire exploit. This is a lie, because someone CANNOT claim immunity to variants that are not YET known.
Logic.
What's the point of speculating? You're just blindly guessing and probably only doing this because you have an Intel CPU.
Get a dictionary. Seriously. Look up "at least" - "at least" means "bare minimum" - if a game used "at least" eight threads, that would mean it either can't use MORE than...or can't use LESS than eight threads, depending on the context of the sentence. No; it can use "bare minimum" one thread, just like every other program.
Yes ACU will use eight threads if your CPU has eight threads and the rest of the PC can keep up. If the PC has twelve threads and the rest of the PC can keep up, it might use all twelve. What's the problem and why state the obvious?
Let's say there's a Single Core, and it's equal to 1000. An equivalent Dual Core, is 500+500. An equivalent Quad Core is 250+250+250+250. What we really care about is threads, though, and we determine one core as equal to two threads.
This is what you said that is wrong. You need more than a 2 x 500 CPU to equal a 1000 CPU.
A benchmark game is a game that attempts to make full use of known hardware/software. It's not that hard of a concept.
My point was that you seem to think that benchmark (or by your terminology, "demanding") games represent the entire game industry, and possibly even other programs; they don't. Just because one game can hyperthread on eight threads, doesn't mean they all can; Vindictus certainly can't. What you seem to think is that Vindictus, or even games that can hyperthread on, say, four threads, can hyperthread on thirty-two threads. No; that's not how it works.
I never said AAA games represent the entire game industry and never said Vindictus can thread to eight threads. They get benchmarked because they are demanding, so if GPU A can run demanding Game N at FPS Y, GPU A can probably run less demanding Game O at FPS Z. Reviewers don't have all day to benchmark every game. However, what do you suggest? You want professional hardware reviewers to start using games such as Vindictus to benchmark GTX 1080/Ti or Vega cards?
But just because ONE game can do that, doesn't mean they ALL can. A game can hyperthread a certain number of threads depending on how it's made. A game that can hyperthread only two threads can't hyperthread eight; a game that can't hyperthread at all, like Vindictus, certainly can't hyperthread eight, and especially not thirty-two.
Having thirty-two threads is only good if any program can operate fully on a single thread of those thirty-two, because otherwise you're limiting yourself to programs that can operate on a large number of threads. By contrast, a core with two threads can run any game with or without hyperthreading just fine; there's less ability to use Amdahl's Law, but more space in each thread, which means you're not compromising the number of processes that can be run.
Yeah I never said all games can. I don't think I said even ACU can. I don't the point of these two paragraphs.
Here we go, same guy but he retested it with one core + Hyper-threading. Spoiler: It's still quite bad. You also need to remember this is a Skylake core overclocked to 4.5GHz.
This is MY argument, not yours. Literally this is my whole argument - that a thirty-two thread CPU won't run Vindictus well at all.
It depends. Definitely not as good as an i7 8700k, but it won't be literally unplayable. A Threadripper 1950X should run Vindictus better than any non-overclocked Ivy Bridge CPU.
This is ALSO my argument. Consumer applications aren't designed for a CPU with thirty-two threads. Now you're just arguing against yourself.
Consumer hardware doesn't have eight threads per core because consumer applications are designed to run on one-two threads per core. 16 core and 32 threaded CPUs are fine. 4 core and 32 threaded CPUs are not fine. Please state where I mentioned thirty-two, which is completely irrelevant to this argument.
All you're doing is misleading the OP and other users in your attempt to promote AMD, a company that has repeatedly lied, cheated, and cheapened over the years. And I've provided every single explanation necessary to show how much its hardware is a gimmick.
If all CPUs work the same, please explain why AMD think they are so confident in their CPUs unlike Intel. I'm not a CPU engineer but if you think you have the knowledge to know why they are, feel free to explain in detail.
It's called marketing and damage mitigation. If they admitted that their CPUs were affected and didn't downplay the issue, their sales would drop. Intel doesn't have to worry about that because it is literally in everything, but AMD has yet to prove itself to the public sufficiently enough for its brand to not be harmed by the situation. It's just business.
What's the point of speculating? You're just blindly guessing and probably only doing this because you have an Intel CPU.
It's not speculation. New variants will be discovered. That's how technology works.
This is what you said that is wrong. You need more than a 2 x 500 CPU to equal a 1000 CPU.
..... What.
Here we go, same guy but he retested it with one core + Hyper-threading. Spoiler: It's still quite bad.
"I am shocked by how much of a difference hyperthreading is making in all of these titles."
Where does he say it's still "quite bad"? It lags, but he's also going pure CPU - no GPU to help alleviate anything. It's true that games like that benefit from more threads but, again, that's not my argument - my argument is that thirty-two threads is bad for general programs and games that aren't designed around hyperthreading, because there's too many cores/threads.
In any case, I don't have anything against AMD. Its hardware is good. Really, really good. I use Intel and Nvidia because they're best for my wallet; high-end chips in my non-custom laptops are not priority for me. I can play any game without any lag or latency (minus peer 2 peer connections), so I'm good. I can even play games like Skyrim and GTA V perfectly smooth on max settings. And all I have is an ASUS G laptop.
So, again, I don't have anything against AMD hardware. My problem with the company is that it's blowing off the issue, which gives me a danger signal that they're not going to properly fix the issue in the products that they currently have set to launch. When a problem this big occurs and a company does not indefinitely delay a release to buckle down and make sure the exploits are fully protected against, that tells me said company shouldn't be trusted.
I would rather wait for Nvidia to make its own APU, or if not just keep getting Intel/Nvidia setups, than gamble with a company that isn't in red alert right now.
It's called marketing and damage mitigation. If they admitted that their CPUs were affected and didn't downplay the issue, their sales would drop. Intel doesn't have to worry about that because it is literally in everything, but AMD has yet to prove itself to the public sufficiently enough for its brand to not be harmed by the situation. It's just business.
That's your opinion. There is no wrong opinion so I'll leave it at that.
It's not speculation. New variants will be discovered. That's how technology works.
Pure speculation. I think the graphics cards released by Nvidia in 2020 will have 6-32GB VRAM because that's how technology works. I have nothing to back it up because I'm speculating.
"I am shocked by how much of a difference hyperthreading is making in all of these titles."
Where does he say it's still "quite bad"? It lags, but he's also going pure CPU - no GPU to help alleviate anything. It's true that games like that benefit from more threads but, again, that's not my argument - my argument is that thirty-two threads is bad for general programs and games that aren't designed around hyperthreading, because there's too many cores/threads.
I never said Hyper-threading is useless. I said single core CPUs are not suitable for 2018 gaming while you insist
Single Core meets all of our needs, and will met all of our needs, regardless of whether we're playing Stardew Valley or Crysis 20,000.
Here is what happens when you try to run Crysis 3 with a dual-core 4.5GHz G3258.
Yes, it even bottlenecks a GTX 750 Ti in this game.
Source: @ 2:07
If all you do in GTA 5 is flying planes, you're fine. But as soon as you start driving as seen at 2:01, "This is pretty bad,".
@2:12
Graphs speak for themselves.
but he's also going pure CPU - no GPU to help alleviate anything.
I can't believe you still think this is how a computer works.
So, again, I don't have anything against AMD hardware. My problem with the company is that it's blowing off the issue, which gives me a danger signal that they're not going to properly fix the issue in the products that they currently have set to launch. When a problem this big occurs and a company does not indefinitely delay a release to buckle down and make sure the exploits are fully protected against, that tells me said company shouldn't be trusted.
I guess this is why Intel will always make money. Because people like you will buy their products no matter what.
All CPUs are affected? Buy Intel. Nothing wrong here.
Non-Intel CPUs are hit the hardest? Buy Intel. The right choice here.
Intel CPUs are hit the hardest? Buy Intel because their competitor must be lying.
Maybe AMD got lucky or they were secretly geniuses, what matters is the result.
Intel, Nvidia and AMD are all far from perfect. But Nvidia and AMD are nowhere close to Intel's greed.
Yes, that's nice. I'm aware all three companies have good products. If you really think 32 threads are useless, please be aware Intel themselves have Xeon CPUs that have 56 threads each. 8 of them can be installed on one motherboard if desired, for a total of 448 threads. If Intel thinks there is a market for that many threads, please tell me why you think AMD is trying to fool people with their 32 threads. By the way, they have server CPUs that have 32 cores and 64 threads to compete with Xeons, however, only two of them can be installed on one motherboard for a total of 64 cores and 128 threads.
Comments
You need to ****ing read. I've already said this before, but you don't seem to understand - known variants =/= all variants. One variant of Meltdown is not indicative of all variants of Meltdown. AMD used what little good news they could to trick the public into believing the lie that it's Meltdown immune; the ONLY tests done on it were for one variant that the testers themselves said was incomplete, as they had yet to do more precise testing.
The exception is not the rule.
Ryzen is not magic.
And to be honest, I don't feel like getting into why more than 8 threads is a gimmick, so here: https://www.makeuseof.com/tag/is-an-octa-core-better-than-a-quad-core-not-always-android-processors-explained/
Hopefully you'll actually read the article this time. But if not, just ask literally anyone else that's talked about the cores before on these forums: More does not mean better. If a program is not designed to make use of multiple threads, then it won't. Most programs don't even efficiently use quad core.
Literally the ONLY reason you'd get a quad, octo, whatever core is if each thread has the capacity to handle any single-threaded program in a single thread without problems. That way you can just separate tons of programs into tons of threads without worry, and everything's better organized. But if you think a processor with thirty-two threads can handle that, you are really puffing the dragon.
Those articles stated that every single CPU is affected by Spectre, and anyone with common sense would know that Meltdown can have new variants that affect all CPUs as well. Not just Intel's; everybody's.
All of the companies have also said they're re-designing future chips to not have the issue. Future chips, as in not any chips that are currently set to launch. Especially since they don't know how they'll redesign the chips yet to not have the exploits.
It seems that you need to go back and read the articles, too, because that information is provided in almost every single one of them.
I seriously don't even know why I'm wasting my time with people who aren't willing to do the research. Go ahead and buy the newest super fancy Ryzen and deal with the dozens or hundreds of variants of the exploits, if you're so certain it'll be immune.
I'll sum it up for programmers to understand the main part (side-channel part) of the Spectre/Meltdown somewhat more easily (it took me a few hours to understand):
This part of the exploit code uses two arrays, one for the out-of-bounds access (small array with like 16 elements) and the other one is used for testing the indexes after (large array with like 256 * 512 elements). It needs a large array because the value of the protected memory address will be copied into it as an index by the speculative execution behaviour of the CPU that will discard this memory content from the registers after it refuses giving access to it for the application. The large array however still holds a memory address that is unreachable for the application but it doesn't need it because the memory content is "stored" as an index of the large array. I quoted it because it's not really stored, it just still holds the protected memory address in the cache (thus faster read time). By probing (dereferencing) each index of the array and measuring the time of each probe you can retrieve the byte it holds from the array index. (so the key part is the large array's index)
Example code from exploit-db.com:
// array1 is the small array // array2 is the large array that holds the value // x is the out-of-bounds offset if (x < array1_size) { temp &= array2[array1[x] * 512]; }This is a real threat however there are no malwares specialized for this so no need to worry yet. And it's quite difficult to fix completely or without heavy performance losses.
1 + 2 = 3
Out of the 5+ tech forums I browse, you're actually the only one to think this. You accuse others of advertising AMD, but now I find it difficult to believe you're not being paid by Intel. If you think there are more than three variants, please do list them. Once again maybe there are six variants or even more but we are only aware of three so we'll just have to accept it and move on. All shall be revealed soon enough anyway.
Assuming the same architecture and clock speed, more cores are better. 6 Coffeelake cores > 4 Coffeelake cores. Really not that hard. That article is from 2015 and for mobile devices, we're in 2018 and on x86 hardware so it's completely irrelevant. Some people use their computer for more than games, stuff that can scale to 32 threads and even more.
We are indeed in the multithreaded era, here is an example of Crysis 3, a 2013 game.
i7 8700k: 6/12
i5 8600k: 6/6
i7 7700k: 4/8
i5 7600k: 4/4
Source:
The quad core i5 still delivers a decent experience, but it doesn't change the fact that practically every AAA game you buy today uses at least 6 threads. I know from personal experience every Assassin's Creed game since ACU uses at least 8 threads.
Well, one company said they felt Meltdown - or rather, the specific part of the problem that allows for the Meltdown exploit - slowed down their game's performance, as the "patch" they released specifically for their game showed better performance afterward, though exactly how much is debatable.
Don't ask me how. I haven't seen their code, and I likely wouldn't understand it.
"soon"
No. That's how this works.
I stated, again and again omg, that there are three KNOOOOOOOOOOOOOOOOOOOOOWWWWWWWWWWWWN variants. KNOOOOOOOOOOOOOOOOOOOOOWWWWWWWWWN. How hard is that to understand? It's a simple ****ing word.
But sure, accept it and move on while AMD fails to mention the 20+ new variants in the next year that will allow people to spy on the chip that it won't fix.
Crysis and Assassin's Creed are benchmark games. Even among AAA games, they're benchmarks. And don't give me that "at least 8 threads" - no. It uses at least ONE thread, and can potentially multi-thread beyond eight. You don't seem to understand how hyperthreading works.
Let's say there's a Single Core, and it's equal to 1000. An equivalent Dual Core, is 500+500. An equivalent Quad Core is 250+250+250+250. What we really care about is threads, though, and we determine one core as equal to two threads.
Now, hyperthreading is super awesome, but it doesn't care how many threads there are, so long as there's more than one. There is no performance difference overall regardless of whether the threads are 500+500 or 125+125+125+125+125+125+125+125. There's more threads to put stuff onto, sure, but all that matters is that we can use hyperthreading properly at all.
With this in mind, understand that Quad Core is already a gimmick. Single Core meets all of our needs, and will met all of our needs, regardless of whether we're playing Stardew Valley or Crysis 20,000. The only exception is games like Vindictus, which are meant to run on Single Core because they take up so much.
Then again, Vindictus can run just fine in a single thread out of my eight threads, because each thread can handle that much. Vindictus can run on 125. That's great.
But you seem to think that not only l33t new AAA games, but even old games like Vindictus will work fine on a thirty-two thread chip. This means you think, out of the sum total of 1000, and the equation of 31.25+31.25+31.25+31.25+31.25+31.25+31.25+31.25+31.25+31.25+31.25+31.25+31.25+31.25+31.25+31.25+31.25+31.25+31.25+31.25+31.25+31.25+31.25+31.25+31.25+31.25+31.25+31.25+31.25+31.25+31.25+31.25, that Vindictus can run on just 31.25.
No. And that you think your 8+ AC game will be able to smoothly play on an equivalent 32 is even more ridiculous.
While I highly doubt Ryzen will let any of its threads only handle that small of an amount, it doesn't change the fact that even IF a single thread on Ryzen could handle 200, its competitors with single core - if they chose to have two threads on that core to allow for hyperthreading - would have 3200+3200 aka 3200 per thread which could handle any number of programs regardless of whether those programs can hyperthread or not.
Increasing the number of threads does not magically increase performance. Hyperthreading itself - just hyperthreading - increases performance.
Wow, you learned something new today!
Oh, and a core can have more than two threads on it. Mind=blown, right? So Quad Cores could totally have thirty-two threads...but they don't, because it's a terrible gimmick.
Rezi doesn't seem to deal with facts... Even though you made a very easy to understand post I wouldn't waste your time with someone that doesn't want to hear the facts if it doesn't side with their opinion. Let them believe what they want there is plenty of information out about it disputing Rezis misguided information.
Oh and last time I checked amd has had processors out for a very long time and they don't suffer from the meltdown bug.. so what's with the premature stuff??
@shaadowcity2
I wouldn't bother replying to this thread anymore. Any intelligent being reading this post from the beginning will understand the message. Going back and reading all the comments, i realized Rezi replied " for a thread asking for a suggestion on which CPU to buy. I am starting to believe this person doesn't actually have knowledge of CPU manufacturers. Also, sounds like they believe Google makes CPUs.
And after reading their recent posts, it seems they don't have any kind of degree related to computer science nor is knowledgable in the subject area.
They do not seem to understand why AMD is fundamentally IMMUNE to meltdown exploit. The way AMD operates is different and AMD is IMMUNE to all variants of meltdown. Perhaps another different unknown exploit exists for AMD CPUs, however worrying about an exploit yet to be discovered is like worrying about World War 3. It's no use trying to explain to them; they won't understand/accept it.
Yes there are three known variants, every tech forum I visit agrees with this. However, you seem to think there are more than three and if you are really think this is the case, please tell us because I'm sure everyone will be very interested. But as far as I can tell, you're making this up due to the fact Intel is more screwed than AMD at the moment.
They get benchmarked because they're demanding. They get benchmarked because people might need to upgrade their PC hard to play it. They get benchmarked because they show the difference in performance of PC hardware. Crysis 3 is chosen by DigitalFoundry because it scales very well with extra threads. They get benchmarked because people actually play these games. Maybe not Crysis 3, but games such as Witcher 3, GTA 5, BF1 and ACO are popular, scale well and demanding. What do you suggest, benchmark Minesweeper?
Of course it uses at least one thread, what doesn't? Yes, ACU uses at least eight threads.
Completely wrong. Read more here https://en.wikipedia.org/wiki/Amdahl's_law
TL;DR: Diminishing returns.
Here is what happens when GTA 5 is ran on a single core CPU:
@3:17
The built-in GTA 5 benchmark didn't even finish and Ashes of the Benchmark averaged less than 1 FPS. I'd say a single core CPU does not meet all our needs.
Where did I say this? Regardless, I'm fairly certain Vindictus runs fine on a Threadriper 1950X. Probably not as a good as an i7 8700k, but it should be good enough. Also games that thread well such as ACU run quite nicely on a Threadripper 1950X.
If a Threadripper 1950X can push a GTX 1080 Ti to 95%+ usage, I'd say that's pretty good.
It depends on the application, but Hyper-threading is definitely not some kind of magic that increase performance for all applications, especially single core applications.
Edit: Here are the results of what happens when I disable SMT of a Ryzen 7 1700 CPU.
As expected, single core performance is unaffected. I would have done it with an Intel CPU, but I don't have a Hyper-threaded Intel CPU available at the moment.
Consumer CPUs don't have eight threads per core because consumer applications aren't designed for them. Not a gimmick, these CPUs exist for a reason.
The problem is that speculative execution doesn't check for memory access violations to speed up the branch prediction. And because it's a speculation, it just executes the predicted code not being sure in any results, because the main execution is still waiting for memory reads because of a prior cache flush, so it can continue after the memory contents are loaded into the cache.
Processors would be much slower without this mechanism so i don't know how AMD solved this problem but reading from protected memory is probably a threat to AMD processors as well. The only processors invulnerable to these are the ones that don't utilize speculative execution, so they don't execute code predicted by the branch predictor unit, they only load the instructions. For example many Raspberry Pi series like the Raspberry Pi 3 Model B that uses an ARM Cortex-A53.
My system:
Intel i7 8800k (delidded) @ 5,1 ghz
Nvidia Geforce 1070
I still have to tune down some of the ingame video setting to maintain 60fps despite having a monster overclock. Usally fps dips in raids or crowded city although cpu nor gpu being used that much. Mostly between 20% - 70% across all cores. Same game with the gpu. The optimization is just bad.
If you wanna go for an amd cpu, pick Ryzen 7 1800x and overclock the hell out of it. A mediocre gpu is enough for Vindictus. Just my 2 cents.
The OP asked which AMD processor to get. Just processor. Then he stated he has a gtx already, which is a GPU. So I figured he was the one confusing CPU and GPU and was actually looking for a GPU.
All CPUs work the same. Your claim is as ridiculous as if one uses analog radio waves and the other uses binary code. All CPUs have company-unique architecture, but all of them function in the same manner that allows for both of these exploits. I've repeated this multiple times, and even linked articles that explained the situation perfectly.
Okay, you apparently don't seem to understand what the word "known" means. You need to get a dictionary.
Known means found. Only three variants have been found, only one for meltdown.
But more CAN be found, the UNKNOWN variants. Immunity to a single variant does NOT represent immunity to ALL variants.
AMD is immune to a SINGLE variant. Yet it claims to be immune to the entire exploit. This is a lie, because someone CANNOT claim immunity to variants that are not YET known.
Logic.
A benchmark game is a game that attempts to make full use of known hardware/software. It's not that hard of a concept.
My point was that you seem to think that benchmark (or by your terminology, "demanding") games represent the entire game industry, and possibly even other programs; they don't. Just because one game can hyperthread on eight threads, doesn't mean they all can; Vindictus certainly can't. What you seem to think is that Vindictus, or even games that can hyperthread on, say, four threads, can hyperthread on thirty-two threads. No; that's not how it works.
Get a dictionary. Seriously. Look up "at least" - "at least" means "bare minimum" - if a game used "at least" eight threads, that would mean it either can't use MORE than...or can't use LESS than eight threads, depending on the context of the sentence. No; it can use "bare minimum" one thread, just like every other program.
I never argued against this. You didn't pay attention to what I said.
It is possible for some AAAAAAAA game to hyperthread through thirty-two threads. The whole point of multiple threads is, of course, to increase performance by decreasing stress a la Amdahl's Law.
But just because ONE game can do that, doesn't mean they ALL can. A game can hyperthread a certain number of threads depending on how it's made. A game that can hyperthread only two threads can't hyperthread eight; a game that can't hyperthread at all, like Vindictus, certainly can't hyperthread eight, and especially not thirty-two.
Having thirty-two threads is only good if any program can operate fully on a single thread of those thirty-two, because otherwise you're limiting yourself to programs that can operate on a large number of threads. By contrast, a core with two threads can run any game with or without hyperthreading just fine; there's less ability to use Amdahl's Law, but more space in each thread, which means you're not compromising the number of processes that can be run.
You should have watched your own video. Single Core =/= single thread. The guy intentionally disabled hyperthreading, even though you can use hyperthreading with just two threads, which most single cores now have.
The video is wrong, because the test is done incorrectly. GTA V has trouble without hyperthreading, of course because hyperthreading increases performance. But the tester disabled hyperthreading despite that being completely unnecessary, which made the audience THINK that the performance hit was due to only having a single core, when in fact it was because the tester forced the game to run on a single thread.
This is MY argument, not yours. Literally this is my whole argument - that a thirty-two thread CPU won't run Vindictus well at all.
This is ALSO my argument. Consumer applications aren't designed for a CPU with thirty-two threads. Now you're just arguing against yourself.
At this point I just think you're trolling. You link things without reading/watching them. You are even trying to argue points against me...that I've already argued against you. You're failing to use basic logic, and ignoring every well-explained point I'm putting across.
All you're doing is misleading the OP and other users in your attempt to promote AMD, a company that has repeatedly lied, cheated, and cheapened over the years. And I've provided every single explanation necessary to show how much its hardware is a gimmick.
Hopefully the OP will be smart enough to check how much each thread can handle, instead of just going after the number of threads. If not, it's his loss.
I just tested it through Steam, and it actually decreased performance for me, but it might be different for someone else.
If all CPUs work the same, please explain why AMD think they are so confident in their CPUs unlike Intel. I'm not a CPU engineer but if you think you have the knowledge to know why they are, feel free to explain in detail.
What's the point of speculating? You're just blindly guessing and probably only doing this because you have an Intel CPU.
Yes ACU will use eight threads if your CPU has eight threads and the rest of the PC can keep up. If the PC has twelve threads and the rest of the PC can keep up, it might use all twelve. What's the problem and why state the obvious?
This is what you said that is wrong. You need more than a 2 x 500 CPU to equal a 1000 CPU.
I never said AAA games represent the entire game industry and never said Vindictus can thread to eight threads. They get benchmarked because they are demanding, so if GPU A can run demanding Game N at FPS Y, GPU A can probably run less demanding Game O at FPS Z. Reviewers don't have all day to benchmark every game. However, what do you suggest? You want professional hardware reviewers to start using games such as Vindictus to benchmark GTX 1080/Ti or Vega cards?
Yeah I never said all games can. I don't think I said even ACU can. I don't the point of these two paragraphs.
Here we go, same guy but he retested it with one core + Hyper-threading. Spoiler: It's still quite bad. You also need to remember this is a Skylake core overclocked to 4.5GHz.
It depends. Definitely not as good as an i7 8700k, but it won't be literally unplayable. A Threadripper 1950X should run Vindictus better than any non-overclocked Ivy Bridge CPU.
Consumer hardware doesn't have eight threads per core because consumer applications are designed to run on one-two threads per core. 16 core and 32 threaded CPUs are fine. 4 core and 32 threaded CPUs are not fine. Please state where I mentioned thirty-two, which is completely irrelevant to this argument.
Prove it.
It's called marketing and damage mitigation. If they admitted that their CPUs were affected and didn't downplay the issue, their sales would drop. Intel doesn't have to worry about that because it is literally in everything, but AMD has yet to prove itself to the public sufficiently enough for its brand to not be harmed by the situation. It's just business.
It's not speculation. New variants will be discovered. That's how technology works.
..... What.
"I am shocked by how much of a difference hyperthreading is making in all of these titles."
Where does he say it's still "quite bad"? It lags, but he's also going pure CPU - no GPU to help alleviate anything. It's true that games like that benefit from more threads but, again, that's not my argument - my argument is that thirty-two threads is bad for general programs and games that aren't designed around hyperthreading, because there's too many cores/threads.
In any case, I don't have anything against AMD. Its hardware is good. Really, really good. I use Intel and Nvidia because they're best for my wallet; high-end chips in my non-custom laptops are not priority for me. I can play any game without any lag or latency (minus peer 2 peer connections), so I'm good. I can even play games like Skyrim and GTA V perfectly smooth on max settings. And all I have is an ASUS G laptop.
So, again, I don't have anything against AMD hardware. My problem with the company is that it's blowing off the issue, which gives me a danger signal that they're not going to properly fix the issue in the products that they currently have set to launch. When a problem this big occurs and a company does not indefinitely delay a release to buckle down and make sure the exploits are fully protected against, that tells me said company shouldn't be trusted.
I would rather wait for Nvidia to make its own APU, or if not just keep getting Intel/Nvidia setups, than gamble with a company that isn't in red alert right now.
And speaking of which, while AMD is trying to fool people with its thirty-two threads, Intel is doing this: https://iq.intel.com/smart-farm-equipment-helps-feed-world/ and Nvidia is doing this: http://mashable.com/2017/12/08/nvidia-releases-most-powerful-gpu/#bXIKqxXc3iq4
So you tell me why I should back a gimmick when I could back tech from companies like that.
That's your opinion. There is no wrong opinion so I'll leave it at that.
Pure speculation. I think the graphics cards released by Nvidia in 2020 will have 6-32GB VRAM because that's how technology works. I have nothing to back it up because I'm speculating.
I never said Hyper-threading is useless. I said single core CPUs are not suitable for 2018 gaming while you insist
Here is what happens when you try to run Crysis 3 with a dual-core 4.5GHz G3258.
Yes, it even bottlenecks a GTX 750 Ti in this game.
Source:
If all you do in GTA 5 is flying planes, you're fine. But as soon as you start driving as seen at 2:01, "This is pretty bad,".
@2:12
Graphs speak for themselves.
I can't believe you still think this is how a computer works.
I guess this is why Intel will always make money. Because people like you will buy their products no matter what.
All CPUs are affected? Buy Intel. Nothing wrong here.
Non-Intel CPUs are hit the hardest? Buy Intel. The right choice here.
Intel CPUs are hit the hardest? Buy Intel because their competitor must be lying.
Maybe AMD got lucky or they were secretly geniuses, what matters is the result.
Intel, Nvidia and AMD are all far from perfect. But Nvidia and AMD are nowhere close to Intel's greed.
https://www.techpowerup.com/240283/intel-released-coffee-lake-knowing-it-was-vulnerable-to-spectre-and-meltdown
Yes, that's nice. I'm aware all three companies have good products. If you really think 32 threads are useless, please be aware Intel themselves have Xeon CPUs that have 56 threads each. 8 of them can be installed on one motherboard if desired, for a total of 448 threads. If Intel thinks there is a market for that many threads, please tell me why you think AMD is trying to fool people with their 32 threads. By the way, they have server CPUs that have 32 cores and 64 threads to compete with Xeons, however, only two of them can be installed on one motherboard for a total of 64 cores and 128 threads.
https://en.wikipedia.org/wiki/List_of_Intel_Xeon_microprocessors#Xeon_Platinum_(octa_processor)
You must give your money to them? What are they, a charity? Buy what's good for you. You're just a number to them.