16 threads AMD CPU?

[CliffTerios]

I feel stupid asking this but I heard about an intel CPU that has 8 cores and hyperthreads up to 16 threads. Is there an AMD equivalent? For consumers, of course.

 

In addition to that, would I benefit from 16 threads when rendering videos? I would assume so?

[themaniac]

nope not yet

[Nineshadow]

Nope.

[pjm0925]

Some of the Opteron server chips have 16/32 cores.  No clue on how those perform, they are pretty old.

[Bob Jim]

Don't get any AMD CPU, other than APUs for low end HTPCs/Casual Gaming builds, and the x4 860k for pairing with a 370/750ti. Also maybe the fx6300 for cheap rendering builds. Anything above that is NOT worth it at all.

[DarkBlade2117]

Avoid AMD until Zen. Then lets see how it works for them. For Intel go for 5820k or 6700k. They will,perform better then any AMD chip out right now in any benchmark.

No there is no equivalent.

[The Official Czex]

I feel stupid asking this but I heard about an intel CPU that has 8 cores and hyperthreads up to 16 threads. Is there an AMD equivalent? For consumers, of course.

 

In addition to that, would I benefit from 16 threads when rendering videos? I would assume so?

 

Nope. Only 8 physical cores with no hyper-threading.

[minibois]

Nah, no consumer 16 thread chips.

And you wont see huge benefits in a 5960X over a 5820k

[Naijin]

There are opteron server grade chips. Wait for Zen architecture in the near future, it might be featuring more than 8 cores with high performance.

[don_svetlio]

Some of the older Opterons had 16 cores I believe

[JefferyD90]

I feel stupid asking this but I heard about an intel CPU that has 8 cores and hyperthreads up to 16 threads. Is there an AMD equivalent? For consumers, of course.

 

In addition to that, would I benefit from 16 threads when rendering videos? I would assume so?

AMD droped HT (Hyper Terminal) technology back in 2007 or 2008.  It actually decreases performance in almost every scenario.

[MageTank]

AMD droped HT (Hyper Terminal) technology back in 2007 or 2008.  It actually decreases performance in almost every scenario.

Umm, an awful lot is wrong with this sentence. HT in the context of this thread is Hyperthreading, not Hyper Terminal. Not only that, when AMD used it, they called it its original name, SMT or Simultaneous Multi-Threading. AMD is adopting SMT again for Zen, rumored to have 8 cores and 16 threads. The "decreases performance" claim is not entirely true either. HT/SMT adds additional heat, which can prevent the best possible turbo boosts from happening, which is where most people see the loss in performance. The difference between an i5 and i7 clocked exactly the same (even with HT on) makes no performance differences in gaming.

 

[SeanAngelo]

I feel stupid asking this but I heard about an intel CPU that has 8 cores and hyperthreads up to 16 threads. Is there an AMD equivalent? For consumers, of course.

 

In addition to that, would I benefit from 16 threads when rendering videos? I would assume so?

nope, only Intel

[Mr.Dingle]

There is the Opteron 6272  .. but thats a server grade CPU so thats a nope.

[JefferyD90]

Umm, an awful lot is wrong with this sentence. HT in the context of this thread is Hyperthreading, not Hyper Terminal. Not only that, when AMD used it, they called it its original name, SMT or Simultaneous Multi-Threading. AMD is adopting SMT again for Zen, rumored to have 8 cores and 16 threads. The "decreases performance" claim is not entirely true either. HT/SMT adds additional heat, which can prevent the best possible turbo boosts from happening, which is where most people see the loss in performance. The difference between an i5 and i7 clocked exactly the same (even with HT on) makes no performance differences in gaming.

 

Back in the day, yes it acutally did decrease performance.  The reason why AMD ditched it.

 

https://www.youtube.com/watch?v=I4PDoy-mi0A

 

Also, the new Zen cores will support up to 4 threads per core.

[Manny Calavera]

he difference between an i5 and i7 clocked exactly the same (even with HT on) makes no performance differences in gaming.

 

hyperthreading does help, there is not difference because some(most) games with i5 is already gpu bottleneck

[MageTank]

Back in the day, yes it acutally did decrease performance.  The reason why AMD ditched it.

 

https://www.youtube.com/watch?v=I4PDoy-mi0A

 

Also, the new Zen cores will support up to 4 threads per core.

I do not believe you watched that video. That video only further proves my point. The FPS in those tests were within margin of error, and HT actually showed an excellent boost in performance on the 1 core, 2 thread case, over the normal 1 core, 1 thread setup. It did not decrease performance in the past. Those that said it did were testing incorrectly, or did not understand exactly what was going on.

 

Can you point me to where you saw Zen supporting 4-way SMT? I have not seen that anywhere, at least not on the consumer readings so far.

 

 

hyperthreading does help, there is not difference because some(most) games with i5 is already gpu bottleneck

Not exactly true. HT helps in games that need the extra threads. There is a reason the i3 does better than the G3258 in Far Cry 4, Dragons Age Inquisition, etc. Pick any game, and compare a core i5 and core i7 at the exact same clock speeds, with HT turned on for the i7, and you will not see much of a difference at all. The notion that an i5 is a GPU bottleneck is just outright false. Even older Sandy Bridge i5's can fully drive a GTX 980 Ti without being the bottleneck. 

[Manny Calavera]

Not exactly true. HT helps in games that need the extra threads. There is a reason the i3 does better than the G3258 in Far Cry 4, Dragons Age Inquisition, etc. Pick any game, and compare a core i5 and core i7 at the exact same clock speeds, with HT turned on for the i7, and you will not see much of a difference at all. The notion that an i5 is a GPU bottleneck is just outright false. Even older Sandy Bridge i5's can fully drive a GTX 980 Ti without being the bottleneck. 

gpu bottleneck not cpu bottleneck

 

you said it yourself, sandy i5 can fully drive 980ti, hence bottlenecked by gpu, not cpu

just because there is no difference in framerate (very small) does not mean extra threads dont or cant get used.

unlike you, i look at benchmarks with dual 980s. http://gamegpu.ru/test-video-cards/igry-2014-goda-protiv-protsessorov-test-gpu.html

 

sandy i7 is better than haswell i5 at same clock... in games aswell.

and xeon 1231v3 is also better than 4.6 4690k. 

 

todays engines dont only use 8 cores (8350 easy example so you can see diff)... they even use 5820k and benefit. 

 

oh, and just so you know how i5 can even be bottleneck,... 100% load on assasins c unity

also i am very tired of only me posting benches all the time

[Nena Trinity]

Opteron maybe? 

[MageTank]

gpu bottleneck not cpu bottleneck

 

you said it yourself, sandy i5 can fully drive 980ti, hence bottlenecked by gpu, not cpu

just because there is no difference in framerate (very small) does not mean extra threads dont or cant get used.

unlike you, i look at benchmarks with dual 980s. http://gamegpu.ru/test-video-cards/igry-2014-goda-protiv-protsessorov-test-gpu.html

 

sandy i7 is better than haswell i5 at same clock... in games aswell.

and xeon 1231v3 is also better than 4.6 4690k. 

 

todays engines dont only use 8 cores (8350 easy example so you can see diff)... they even use 5820k and benefit. 

 

oh, and just so you know how i5 can even be bottleneck,... 100% load on assasins c unity

also i am very tired of only me posting benches all the time

The structure of your sentence made it unclear exactly which was the bottleneck. Either way, that point is resolved now. Also, do not say "unlike you" because i look at every benchmark, not just single GPU's. Oh, and i never said the extra threads don't get used. I said beyond 4 cores, extra threads become useless. Let's broaden the spectrum a little, shall we?

 

 

Notice how close the i5 4690k, i7 4790k, and i7 5960x are in average frame rate? Sure, every now and then, the 5960x spikes to a 5% frame improvement over the other CPU's, but 5% is not substantial at all, given its massive difference in price. 

 

 

Notice in this video, during all of the games, performance was relatively the same. The only time the 5930k pulled ahead, was in a synthetic benchmark. 

 

 

Same story here. 

 

ACU seems to be the only exception to the rule, if what you linked is truly accurate. I don't have the game, so i can't test it myself, but people have often complained about Ubisoft ports in the past. Battlefield 4 is normally extremely accommodating when it comes to using more cores, and even it showed little difference between the higher core/thread counts in most testing.

 

Point is, the difference it makes is 9 times out of 10, pointless. Plenty of people running ACU on i5's with overclocked 980's and still hitting that 60fps window just fine without any issues. For gaming, 4 cores/4 threads is plenty enough.

[JefferyD90]

I do not believe you watched that video. That video only further proves my point. The FPS in those tests were within margin of error, and HT actually showed an excellent boost in performance on the 1 core, 2 thread case, over the normal 1 core, 1 thread setup. It did not decrease performance in the past. Those that said it did were testing incorrectly, or did not understand exactly what was going on.

 

Can you point me to where you saw Zen supporting 4-way SMT? I have not seen that anywhere, at least not on the consumer readings so far.

 

 

Not exactly true. HT helps in games that need the extra threads. There is a reason the i3 does better than the G3258 in Far Cry 4, Dragons Age Inquisition, etc. Pick any game, and compare a core i5 and core i7 at the exact same clock speeds, with HT turned on for the i7, and you will not see much of a difference at all. The notion that an i5 is a GPU bottleneck is just outright false. Even older Sandy Bridge i5's can fully drive a GTX 980 Ti without being the bottleneck. 

http://www.extremetech.com/extreme/198386-amds-next-gen-cpu-leak-14nm-simultaneous-multithreading-and-ddr4-support Edit: they have removed the part talking about the 4 way threads.  If you do a web search for it you can view this in the cached version of the page.

 

That video was just made back a few years ago when Multi Threads where actually matured quite a bit.  Back in 2004-2006, when multi threads where still new (first off) Windows didn't handle it well, (second) the new advent of dual core CPU's made more threads even harder to manage.  It made sense, at that time, that more threads wasn't exactly how to increase performance.  The problem is AMD never bothered to get back on the multi thread bandwagon.  Once Windows Vista matured, and Windows 7 took over this gap was (essentially) eliminated.  The video that I linked showed a slight decrease in performance once you increase those threads past 4 (if I remember right) and this is in a era when threads were much more mature.  This was meant to be more of "Here's what AMD thought" more so than a "Here's how it actually is".

[MageTank]

http://www.extremetech.com/extreme/198386-amds-next-gen-cpu-leak-14nm-simultaneous-multithreading-and-ddr4-support Edit: they have removed the part talking about the 4 way threads.  If you do a web search for it you can view this in the cached version of the page.

 

That video was just made back a few years ago when Multi Threads where actually matured quite a bit.  Back in 2004-2006, when multi threads where still new (first off) Windows didn't handle it well, (second) the new advent of dual core CPU's made more threads even harder to manage.  It made sense, at that time, that more threads wasn't exactly how to increase performance.  The problem is AMD never bothered to get back on the multi thread bandwagon.  Once Windows Vista matured, and Windows 7 took over this gap was (essentially) eliminated.  The video that I linked showed a slight decrease in performance once you increase those threads past 4 (if I remember right) and this is in a era when threads were much more mature.  This was meant to be more of "Here's what AMD thought" more so than a "Here's how it actually is".

I watched the entire video. What it actually showed was slight to no difference between having more cores/threads. The difference was within a margin of error. It did not show a decrease in performance. Only two games that i know of had issues with HT causing a drop in performance. World of Warcraft and Final Fantasy 13.

 

World of Warcraft had an issue in which it spawned 3 threads. Two on the same core, one on another core. It caused severe stuttering, too many issues. They made a thread to help people on the wiki: http://wowwiki.wikia.com/wiki/CVar_processAffinityMask

 

Final Fantasy 13 has stuttering issues too, which is alleviated by turning HT off. I assume it has the same issue that WoW had back in the day, which is why turning HT off removes the stuttering.

 

https://steamcommunity.com/app/292120/discussions/0/594820656474485336/ 

http://pcgamingwiki.com/wiki/Final_Fantasy_XIII

 

 

 

Disable hyperthreading on Intel processors^[22]

Somewhat extreme.

These seem to be isolated instances, and is a result of the software being broken, not HT itself. As i said before though, an i5 would suffice, as gamers do not need HT.

[Manny Calavera]

The structure of your sentence made it unclear exactly which was the bottleneck. Either way, that point is resolved now. Also, do not say "unlike you" because i look at every benchmark, not just single GPU's. Oh, and i never said the extra threads don't get used. I said beyond 4 cores, extra threads become useless. Let's broaden the spectrum a little, shall we?

 

 

Notice how close the i5 4690k, i7 4790k, and i7 5960x are in average frame rate? Sure, every now and then, the 5960x spikes to a 5% frame improvement over the other CPU's, but 5% is not substantial at all, given its massive difference in price. 

 

 

Notice in this video, during all of the games, performance was relatively the same. The only time the 5930k pulled ahead, was in a synthetic benchmark. 

 

 

Same story here. 

 

ACU seems to be the only exception to the rule, if what you linked is truly accurate. I don't have the game, so i can't test it myself, but people have often complained about Ubisoft ports in the past. Battlefield 4 is normally extremely accommodating when it comes to using more cores, and even it showed little difference between the higher core/thread counts in most testing.

 

Point is, the difference it makes is 9 times out of 10, pointless. Plenty of people running ACU on i5's with overclocked 980's and still hitting that 60fps window just fine without any issues. For gaming, 4 cores/4 threads is plenty enough.

thanks for benches

so basicly, 4 cores clocked as much as possible, but 4c8t if there is too much work to do will benefit

 

thanks m8 ! Will check more benches, but tommorow, i will check OC benches on pclab and techno kitchen.

Btw, watch dogs also likes 4.9 i7.

[MageTank]

thanks for benches

 

so basicly, 4 cores clocked as much as possible, but 4c8t if there is too much work to do will benefit

 

thanks m8 ! Will check more benches, but tommorow, i will check OC benches on pclab and techno kitchen.

Btw, watch dogs also likes 4.9 i7.

Watch Dogs is also another Ubisoft game. Ubisoft's motto is: "Why should we optimize it when you can just throw more hardware at it?"

 

I'll have to go out and buy ACU and do some tests myself. After you showed me it scaling nearly perfectly with 8 threads, i have some things i really want to test.

[Sintezza]

not sure why the whole intel vs amd debate for gaming gets broaght up to this thread.

But to aswer the original question, yes there are AMD opteron server grade chips with more cores.

 

also currious if topic starter is even going to look back to his thread.

Or if this thread is another intel vs amd flame war troll thread.

 

Fingers crossed