FX 8370 Review

[monstercameron]

Fp is so bad on fx bd, pd and sr but its all in the name of fus..hsa. I kinda hope AMD doubles down on their fx design methodology. More fast integer, decent St fpu and delegate the rest of the fp workload to the gpu.

 

[frozentundra123456]

And for someone who's gaming is not the primary goal???? Why talk about gaming only? If gaming is all you do then yes don't pick FX, but if you need your PC for other stuff that might benefit from several threads too, FX is a very compelling option.

That statement was in direct response to a gaming benchmark.

 

[ShintaiDK]

Holding the Stilt's comments about Vishera-k aside, the 8370SE does not surprise here much. Nothing to see here, move right along . . .

Now, if the Vishera-k info he posted is legit, then that is a somewhat-interesting development. Could we start seeing 5 ghz overclocks on the 8320SE?

The Stilt? You mean the guy that lacks basic understanding of Tcase? He may get more credit than deserved.

 

[Ramses]

The Stilt? You mean the guy that lacks basic understanding of Tcase? He may get more credit than deserved.

That long bit from stilt does not jive with what I've read in AMD docs and seen explained by some pretty well respected folks before. It's a complex-ish concept so I get why there is confusion.

Edit: http://www.overclock.net/t/1128821/amd-temp-information-and-guide

 

[Ramses]

only that there are more well rounded and efficient offerings at very similar price

I don't think anyone is arguing that really, least I'm not in Sept of 2014.
There are just a ton of other reasons that go into a buying decision.
It isn't black and white is what I hoped to illustrate.
Benchmarks are, but people are not.

 

[MiddleOfTheRoad]

So under Dx an i3 is faster, and even with mantle the i3 is as fast as the 8350 while using 100 watts less power. Doesnt seem like a very good example to justify the 8350.

It does if you plan on using that computer for HD video editing -- which chokes an i3.

 

[MiddleOfTheRoad]

I do admire that they brought 8 cores to 95w, means that you get to pick between the fx6300 and the 8370e as your best 95w options.

Agreed -- the 8370e is probably now the deFacto choice for AMD builds on the World Community Grid.

 

[Abwx]

Fp is so bad on fx bd, pd and sr but its all in the name of fus..hsa. I kinda hope AMD doubles down on their fx design methodology. More fast integer, decent St fpu and delegate the rest of the fp workload to the gpu.

FP perf is good once the soft is correctly optimised for the CPU, check PoVray score on the second graph :

http://www.computerbase.de/2014-09/amd-fx-8370e-im-test/2/#diagramm-pov-ray

That s a long shot from 3DPM, i tested this latter on a Athlon XP, it works and the AXP has the same IPC as a Piledriver or Steamroller in this test, given that it has no SSE2 it s obvious that it s using X87 wich explain the better score of the Phenom wich has better X87 perfs than BDozer.

Edit : The Athlon XP ST score is 31.2 , that is 20% better IPC than the Piledriver and Steamroller and 5% lower IPC than the Phenom II......

 

[nenforcer]

The fact that both consoles, the XBOX One and Sony Playsation 4, have 8-core AMD Jaguar CPU's makes me believe these FX 8-core chips will have a really long life time in terms of gaming.

Most PC ports, starting with Grand Theft Auto V, should scale and perform very competitively on the AMD FX chips for the life of these new consoles (until 2020ish).

 

[cytg111]

The fact that both consoles, the XBOX One and Sony Playsation 4, have 8-core AMD Jaguar CPU's makes me believe these FX 8-core chips will have a really long life time in terms of gaming.

Most PC ports, starting with Grand Theft Auto V, should scale and perform very competitively on the AMD FX chips for the life of these new consoles (until 2020ish).

8 cores jaguar, FX or Haswell-E all suffers from the same terminal condition: Diminishing returns.
One. Amdahls law.
Two. Synchronization overhead, the more cores you have to synchronize the bigger the overhead becomes .. That 8'th core will rarely go over 50% efficiency. A tech like Intels TSX would greatly help leverage this issue. Unfortunately they derped the very feature that should help pave the way for the many-core future.

 

[jhu]

8 cores jaguar, FX or Haswell-E all suffers from the same terminal condition: Diminishing returns.
One. Amdahls law.
Two. Synchronization overhead, the more cores you have to synchronize the bigger the overhead becomes .. That 8'th core will rarely go over 50% efficiency. A tech like Intels TSX would greatly help leverage this issue. Unfortunately they derped the very feature that should help pave the way for the many-core future.

Depends on the workload. For workloads that are easily parallelized, the effects of Amdahl's law isn't noticeable at 8 cores on Linux.

 

[SPBHM]

The fact that both consoles, the XBOX One and Sony Playsation 4, have 8-core AMD Jaguar CPU's makes me believe these FX 8-core chips will have a really long life time in terms of gaming.

Most PC ports, starting with Grand Theft Auto V, should scale and perform very competitively on the AMD FX chips for the life of these new consoles (until 2020ish).

consoles run a different version of the game, because of the OS/fixed spec differences

Dead Rising 3 was made as an Xbox One exclusive and released in 2013 (1.75GHz 8 core Jaguar) and runs at 30FPS (with slowdowns)

doesn't seem like it brings much for AMD CPUs

perhaps newer Xbox One games PC ports will be more optimized for MT, but the PC version of dead rising 3 loves 4 threads, like most games from the past 5 years.

best case scenario I can see for the 8 core FX for gaming is to close the gap for some games (like a BF4 or anything highly optimized for more than 4 threads), but for many years to come I think other games will be heavily dependent on the core performance and limited to 1-4 threads, so it's hard to see the FX as a better option for gaming than most Intel CPUs, but let's wait and see, GTA 5 should be a good one.

 

[USER8000]

consoles run a different version of the game, because of the OS/fixed spec differences

Dead Rising 3 was made as an Xbox One exclusive and released in 2013 (1.75GHz 8 core Jaguar) and runs at 30FPS (with slowdowns)

doesn't seem like it brings much for AMD CPUs

perhaps newer Xbox One games PC ports will be more optimized for MT, but the PC version of dead rising 3 loves 4 threads, like most games from the past 5 years.

best case scenario I can see for the 8 core FX for gaming is to close the gap for some games (like a BF4 or anything highly optimized for more than 4 threads), but for many years to come I think other games will be heavily dependent on the core performance and limited to 1-4 threads, so it's hard to see the FX as a better option for gaming than most Intel CPUs, but let's wait and see, GTA 5 should be a good one.

 

[frozentundra123456]

Those are pretty unusual results for a modern game. I have to admit I was pretty shocked when I saw them. It does perfectly illustrate the problem with FX though for gaming, it is wildly inconsistent. In some games it is very competitive, while in others it is atrociously slower. And there are no games in which FX is really marked faster than an i5 or i7. One can blame the software, or speculate what will happen in the future, bur right now, the performance "is what it is", which in equal at best and in some cases, markedly slower, while negating some of the initial cost savings by using more power. This is why I don't understand those who so vigerously defend the higher power consumption of the FX. If the performance were reversed and FX was faster, I would have no problem with the higher power use, but giving up performance while using more power all for something like a 10 percent cost savings based on the total cost of a system, just seems like a poor compromise, especially for a relatively expensive hobby like PC gaming.

 

[USER8000]

Those are pretty unusual results for a modern game. I have to admit I was pretty shocked when I saw them. It does perfectly illustrate the problem with FX though for gaming, it is wildly inconsistent. In some games it is very competitive, while in others it is atrociously slower. And there are no games in which FX is really marked faster than an i5 or i7. One can blame the software, or speculate what will happen in the future, bur right now, the performance "is what it is", which in equal at best and in some cases, markedly slower, while negating some of the initial cost savings by using more power. This is why I don't understand those who so vigerously defend the higher power consumption of the FX. If the performance were reversed and FX was faster, I would have no problem with the higher power use, but giving up performance while using more power all for something like a 10 percent cost savings based on the total cost of a system, just seems like a poor compromise, especially for a relatively expensive hobby like PC gaming.

Because in many countries an FX8320 is the same price as a Core i3?? Or the fact even H series ATX motherboards are the same price as 970 based overclocking boards??

The worst thing is that the vast majority of games will run fine even on an oldish quad core,especially popular ones with millions or tens of millions of sales.

Hardware is very cheap relative to income in the US - not most of the world and hence 99% of games will run fine on older hardware or not the greatest newer hardware.

This is why hardware enthusiasts on forums need to measurebate about the 1%(or less) of games which might need decent hardware to run on,especially when they want to run games at Ultra+ settings with 16X fancy AA and all the stupid settings which add 2% improved image quality at double the processing power.

They need to justify spending more on hardware or constantly upgrading to fulfill their habit.

Look at the amount of people who have upgraded from high clocked SB Core i5 CPUs to Haswell for one excuse or another.

If anything look at Dead Rising:

http://gamegpu.ru/images/remote/htt...t_GPU-Action-Dead_Rising_3-test-dr_3_1920.jpg

It needs at least an HD7950 or Nvidia equivalent to break 60FPS at 1920X1080 without AA at max settings PAIRED to a HIGH END OVERCLOCKED CPU.

An HD7950 or GTX660TI are NOT cheap gaming cards.

I would suspect most people running it would be doing so at lower settings anyway and running it at under 60FPS.

Even then it will be a better experience than the console version.

 

[Sweepr]

So AMD's GPUs scale a lot better with Intel CPUs in Dead Rising 3. Not the first time I saw something like that but it's kinda ironic.

 

[USER8000]

So AMD's GPUs scale a lot better with Intel CPUs in Dead Rising 3. Not the first time I saw something like that but it's kinda ironic.

Yep,since NV threads their standard drivers it appears.

Edit!!

It looks like my Xeon E3 1230 V2 and GTX660TI won't run Dead Rising 3 at 1920X1080 at massive framerates. A faster GTX670 is barely at 30FPS for minimums and 40FPS for average framerates.

Game looks terribly optimised especially for NV GPU owners!

WTF is going on there??

 

[frozentundra123456]

I would interpret that as nVidia gpus taking better advantage of multiple cores relative to amd, since this game does not use mantle.

 

[SPBHM]

An HD7950 or GTX660TI are NOT cheap gaming cards.

I would suspect most people running it would be doing so at lower settings anyway and running it at under 60FPS.

Even then it will be a better experience than the console version.

the cheap 260x is holding 30 min 40 average at 1200P very high, AMD CPUs would most likely drop it more than Intel CPUs compared to this result (look at the 290x CPU scaling)


the CPU test with the Nvidia GPU shows much lower driver overhead I think, but it still shows the game is in love with 4 threads max (compare the 4300 to the 8350),

 

[Phynaz]

So AMD's GPUs scale a lot better with Intel CPUs in Dead Rising 3. Not the first time I saw something like that but it's kinda ironic.

What's ironic is when AMD ships GPU review units installed in Intel systems. Or when they reviewers guides say to use Intel CPUs.

 

[Ramses]

What's ironic is when AMD ships GPU review units installed in Intel systems. Or when they reviewers guides say to use Intel CPUs.

One mans irony is anothers realism.

 

[ShintaiDK]

So AMD's GPUs scale a lot better with Intel CPUs in Dead Rising 3. Not the first time I saw something like that but it's kinda ironic.

Even more ironic that AMD didnt bother to multithread their DX driver. Hence penalizing their own CPUs.

 

[Fox5]

Dead Rising 3 with Phenom II x6 @ 3.3ghz and R9 290.

6 core usage whenever it's loading data, pretty solid 3 to 4 core usage during normal action. Never drops below 30fps during normal action, drops to about 20fps when it's loading data and all 6 cores are pegged, hits 60fps quite frequently.

This is with all settings maxed.
For comparison, the same scene at lowest settings hovers around 50fps instead, so I do seem to be largely cpu limited.
Nice boost on nvidia cards though, the fact that it even boosts dual and quad cores when the game itself is regularly using 3-4 cores really says something about the driver overhead of nvidia's drivers vs AMD... that's a mantle like boost.

 

[Fox5]

Looking at this, the min frame rates (and the averages as well) pretty much double up until you hit the Intel cpus, which get a ~50% boost.

Given that the game is well threaded, this might imply that AMD's drivers really are limiting performance here.

Say dead rising 3 has 6 threads, each doing 6 independent tasks. The primary thread may handle graphics rendering related stuff and directly interact with the DX11 API, and the graphics driver runs on the same core. That could leave 2 threads busy waiting for the primary thread to finish, and 3 additional threads in DR3 that can spin up as needed for non-graphics/physics tasks.
Anyhow, since the workload is the same regardless of gpu in play, it really gives a strong indication that whether through threading or simply being more efficient, nvidia's driver is leaving a lot more time available for that primary thread than AMD's driver.

 

[monstercameron]

Looking at this, the min frame rates (and the averages as well) pretty much double up until you hit the Intel cpus, which get a ~50% boost.

Given that the game is well threaded, this might imply that AMD's drivers really are limiting performance here.

Say dead rising 3 has 6 threads, each doing 6 independent tasks. The primary thread may handle graphics rendering related stuff and directly interact with the DX11 API, and the graphics driver runs on the same core. That could leave 2 threads busy waiting for the primary thread to finish, and 3 additional threads in DR3 that can spin up as needed for non-graphics/physics tasks.
Anyhow, since the workload is the same regardless of gpu in play, it really gives a strong indication that whether through threading or simply being more efficient, nvidia's driver is leaving a lot more time available for that primary thread than AMD's driver.

hmmm am i missing something here?

it seems that amd cards are doing better here. I dont get your bolded claim.