GTA V CPU benches - AMD gets hammered (again)

[escrow4]

*** The VC&G thread isn't really focussed on CPU, this one will be ***

A 9590 cannot crack 60FPS minimums, an 8350 can't crack 50FPS, and those are the minimum dips you will notice and feel. Plus look at Haswell vs SNB there is a difference. The game can use a hexa core although grrr no 5820K. With Haswell's efficiencies it would be interesting to see. Also, note the difference between the slowest Haswell refresh and 4690, 3/4 FPS so if you are buying a locked Haswell, get the Highest clocked one.

 

[Phynaz]

Write off the Pentiums. Minimums in the single digits, ouch!

 

[cytg111]

Write off the Pentiums. Minimums in the single digits, ouch!

Yea, wonder why, they sort of plummets compared to the rest there... cache sizes?

 

[janeuner]

Write off the Pentiums. Minimums in the single digits, ouch!

For real. At least the AMD numbers are playable. Looks more like dual-cores get hammered.

They must have pipelined the logic across 3+ threads, and then blocked on communication between them. That would mostly explain why the i3 benchmarks weren't shit as well.

Profanity is not allowed in the technical forums
Markfw900

Edit:

Spoiler

4 threads, and maybe those are spinlocks? lol?

 

[Flapdrol1337]

Yea, wonder why, they sort of plummets compared to the rest there... cache sizes?

No, the 3MB i3 has no probs.

I think it plummets because the high priority threads hog the 2 cores, so low priority tasks don't get finished in time, then stall the game until they're done.

Fps limiter or playing gpu limited might "cure" it, bf4 is also perfectly smooth with fps limiter.

 

[SPBHM]

I would like to see a CPU test using AMD graphics card.

 

[cytg111]

No, the 3MB i3 has no probs.

I think it plummets because the high priority threads hog the 2 cores, so low priority tasks don't get finished in time, then stall the game until they're done.

Fps limiter or playing gpu limited might "cure" it, bf4 is also perfectly smooth with fps limiter.

Yea, but, well, I suppose.. whats left, however improbable, must be the truth? It warrants further inquiry though.

 

[ZGR]

The numbers seem about right. My OC'd i5 is near 100% load with this game. Makes me think the i5 may not be enough. Hopefully Broadwell-K is coming soon.

I'm getting higher fps than the GTX 970, but that could be due to settings or any number of things.

 

[jpiniero]

I'd like to see them throw in an APU in there and an Intel CPU older than Sandy Bridge.

Yea, wonder why, they sort of plummets compared to the rest there... cache sizes?

I think the game is using 4 main threads actually. The 5960X might be benefiting from the extra cache.

 

[Sweepr]

Interesting results, considering people were expecting FX to hammer Haswell Core i3 and Core i5 in this game. 4-threaded Core i5 4690K beat the power hogs at stock and even 2C/4T Core i3 held its own. Performance per core still matters after all.

 

[Carfax83]

I think the game is using 4 main threads actually. The 5960X might be benefiting from the extra cache.

This. People are ranting and raving about how well threaded the game is, but the benchmarks reflect that the game is using four threads at the most.

Most of the other current gen engines scale to 6 threads..

 

[frozentundra123456]

*** The VC&G thread isn't really focussed on CPU, this one will be ***

A 9590 cannot crack 60FPS minimums, an 8350 can't crack 50FPS, and those are the minimum dips you will notice and feel. Plus look at Haswell vs SNB there is a difference. The game can use a hexa core although grrr no 5820K. With Haswell's efficiencies it would be interesting to see. Also, note the difference between the slowest Haswell refresh and 4690, 3/4 FPS so if you are buying a locked Haswell, get the Highest clocked one.

Well at least the 8350 beats an i3, in the game.gpu benchmarks at least.

Actually, AMD looks somewhat competitive in the game.gpu test, while looking really awful in the PC lab benches.

The two tests dont agree very well either, if you look closely. PC Lab is getting almost as fast framerate with a GTX970 as Game.gpu is getting with 980 SLI. Maybe SLI doesnt scale well yet, or the 980s are running out of vram? Or maybe game.gpu is running Ultra or with some other performance sapping setting turned up.

 

[Abwx]

When looking at the CPu usage the 2600K is loaded at 72.5% for the main cores while the 8350 is at 61%, the 20% higher loading of the former translate in 20% higher Fps, how surprising that a CPU that is underutilised will perform not as well as a one wich is better maxed out....

 

[escrow4]

5930K @ 3.7GHz in game is being utilised, all cores peak at full clocks. No problems with performance here.

 

[TuxDave]

When looking at the CPu usage the 2600K is loaded at 72.5% for the main cores while the 8350 is at 61%, the 20% higher loading of the former translate in 20% higher Fps, how surprising that a CPU that is underutilised will perform not as well as a one wich is better maxed out....

The FPS numbers and the CPU utilization are both symptoms of a problem but the cause is still unknown.

 

[Abwx]

The FPS numbers and the CPU utilization are both symptoms of a problem but the cause is still unknown.

Really dont know, obviously the FX8350 throughput in this game is 89% greater than the FX4300 output, yet the score difference is only 23%..

Must admit that almost twice the computation for a so low improvement let me wondering....

 

[SPBHM]

When looking at the CPu usage the 2600K is loaded at 72.5% for the main cores while the 8350 is at 61%, the 20% higher loading of the former translate in 20% higher Fps, how surprising that a CPU that is underutilised will perform not as well as a one wich is better maxed out....

higher framerate = higher load? different bottlenecks? the i7 is much faster for the main thread, have much faster l3 and so on!?

 

[escrow4]

Really dont know, obviously the FX8350 throughput in this game is 89% greater than the FX4300 output, yet the score difference is only 23%..

Must admit that almost twice the computation for a so low improvement let me wondering....

No FX gets close to 60FPS minimum. Period. AMD fails once again when a basic i5 gets way higher minimums.

 

[Magic Carpet]

@escrow4

The 8350 is a good processor, but for other tasks. Glad, you have the better suited part in your rig. Hope, you are enjoying the game :thumbsup:

 

[Ramses]

Anything more than 40ish is a gift on a CPU this old that wasn't exactly gods gift to gaming when it was new. The "amd gets hammered" thread title is frankly depressing, not like this is a new chip just being tested. It's about as relevant as "Pentium III gets hammered" at this point imo. At this point it's just mean spirited and unkind and not doing anyone any good to keep flogging the proverbial dead horse. The results speak for themselves and they have been for years.

 

[Abwx]

higher framerate = higher load? different bottlenecks? the i7 is much faster for the main thread, have much faster l3 and so on!?

Yes but if the main thread was late in the FX8350 this wouldnt get it to reach such a CPU utilisation, since this thread, and hence core, is distributing the work to other cores any slowness in its execution would yield a low utilisation of the other cores.

Also i find odd that the FX8150 is better than the FX6300 given that Vishera has 13.5% better IPC in games than Zambezi, to end on a critical note on all thoses tests, or rather sites, i find that they are missing a lot of hints by not using a FX4350 as comparison with the FX4300 as the former full 8MB cache (in lieu of 4MB) would provide some infos about games sensitivity to this parameter.

 

[erunion]

Anything more than 40ish is a gift on a CPU this old that wasn't exactly gods gift to gaming when it was new. The "amd gets hammered" thread title is frankly depressing, not like this is a new chip just being tested. It's about as relevant as "Pentium III gets hammered" at this point imo. At this point it's just mean spirited and unkind and not doing anyone any good to keep flogging the proverbial dead horse. The results speak for themselves and they have been for years.

But the FX 8350 isn't a Pentium III, its part of AMDs current line up.

If there wasn't a group of people consistently claiming how great AMD chips are for gaming, you would have a valid point. But there is. Remember that recent thread about how future proof FX will be?

 

[frozentundra123456]

The FPS numbers and the CPU utilization are both symptoms of a problem but the cause is still unknown.

I dont know how accurate those core utilization graphs are, and what they really mean.
I am sure they show the general trend of cpu usage, but I dont know if you can really break it down in too much detail and try to determine anything.

For instance, you can see just the opposite in the Battlefield Hardline cpu utilization chart.
FX cores show higher usage, but it is still slower. And in the same graph one core of the 3970x is utilized at 83% while the max core usage on the 2600k is only 72%. Doesnt make sense that a core in a six core cpu would be utilized higher than the core in a quad one. So I think it is useful to show a general trend, you shouldn't try to infer too much from it.

 

[frozentundra123456]

Anything more than 40ish is a gift on a CPU this old that wasn't exactly gods gift to gaming when it was new. The "amd gets hammered" thread title is frankly depressing, not like this is a new chip just being tested. It's about as relevant as "Pentium III gets hammered" at this point imo. At this point it's just mean spirited and unkind and not doing anyone any good to keep flogging the proverbial dead horse. The results speak for themselves and they have been for years.

Well the pentium III isnt Intel's current top of the line processor, so that is hardly a relevant analogy. The FX, while it may be old, is the best AMD has to offer. So while I agree the title of the thread is a bit harsh, being old is no excuse for the FX, in fact it is more of an indictment of AMD that they have lagged so far behind.

 

[Ramses]

But the FX 8350 isn't a Pentium III, its part of AMDs current line up.

If there wasn't a group of people consistently claiming how great AMD chips are for gaming, you would have a valid point. But there is. Remember that recent thread about how future proof FX will be?

I get what you're saying and I have no excuse for anyone saying it's a great gaming cpu in 2015, but it's old. Everyone knows it's old, and that it trailed Intel stuff gaming when it wasn't old more often than not. But they did give decent performance most of the time and if gaming is only part of what one is into some of aren't a bad deal still at times. It looks like an 8350 would probly gun gta v just fine for most people at 1080p. I had an 8350, a 9590 and now a 4790K, I think I've sampled enough plates recently to say they were a fine cpu for what they were. It's just depressing to see the aggressive beating of a sad dead horse. Churchill said something about it costing nothing to be polite to a man if you had to kill him.