GTA V CPU benches - AMD gets hammered (again)

[Abwx]

Are you looking at pclab's results? <- That should not happen.

I just browsed through their site, they are not bad for everything and have some interesting reviews.

Alas, as displayed by their testbeds pages they are officialy sponsored by Intel, so i guess that they must abide by a few rules to keep receiving some otherwise unaffordable gear, at least they are one of the very few sites that did a review of the A8 7650K...

Hardware.fr that i often use as reference are actualy in the same boat although their reviews are generaly more balanced..

Also it look like the OP non negociable 60 fps mini got his CPU being ultimately hammered as well....

 

[NTMBK]

hmmmm

could be patched in.

my understanding was DX12 is basically a software change in a more efficient rendering pipeline, and basically all the DX11.1 cards are ready to go with DX12.

It's a totally new API paradigm, you would need to rewrite most of the game engine.

 

[RampantAndroid]

hmmmm

could be patched in.

my understanding was DX12 is basically a software change in a more efficient rendering pipeline, and basically all the DX11.1 cards are ready to go with DX12.

Sure. But it isn't a small patch. It's a lot of work on the game dev's part to do it afaik. Maybe something like GTA V would get it, but most games available now or available even within a year of now won't be DX12.

 

[BigDaveX]

It sounds like it's pretty much the same situation we had where DirectX 10 was such a major departure from previous versions that games had to include separate DirectX 10/11 and DirectX 9c codepaths to accommodate all the different possible hardware and software setups. And I don't think there were too many games circa 2006-07 that were released with DirectX 9c renderers and had DirectX 10 ones patched in later.

 

[Yuriman]

hyperthreading has potential to add 20% performance.

dual-core in modules have potential to add 80% performance.

With 5 heavily loaded threads, the chip with the HT is going to show as 'more heavily utilized'

20% might have been true with Sandy Bridge, but with Haswell it seems that 25-35% is more typical, and outliers (mostly scientific computing) gaining 50-70% more performance from HT.

 

[Spjut]

PCgameshardware.de has done a core scaling test, using an i7 5820k downclocked to 2,5 ghz

http://www.pcgameshardware.de/GTA-5...219948/Tests/GTA-5-PC-Test-Benchmark-1156215/

 

[rainy]

Personally, I would like to understand intention or intentions of OP - if his idea was just to create another thread for bashers/trolls, for sure he was successful.

 

[sm625]

It is surprising that the 2500k does so poorly, but then again they didnt run the 2500k at a reasonable clock speed, like 4.4GHz.

 

[BigDaveX]

Personally, I would like to understand intention or intentions of OP - if his idea was just to create another thread for bashers/trolls, for sure he was successful.

There actually were a few people hoping that GTA5 PC would be a game that was more favourable to the FX series, given that it's almost certainly based on the PS4/Xbone version of the game, and those consoles use a CPU set-up a little more similar to FX than Intel's line. Regardless of what the OP's intentions were, it seems that those hopes haven't been borne out.

 

[desprado]

There actually were a few people hoping that GTA5 PC would be a game that was more favourable to the FX series, given that it's almost certainly based on the PS4/Xbone version of the game, and those consoles use a CPU set-up a little more similar to FX than Intel's line. Regardless of what the OP's intentions were, it seems that those hopes haven't been borne out.

only red team was hoping but normal people know what are limits of AMD CPU and that is the reason it is non exist in terms of gaming.

 

[Head1985]

Updated chart with oc sandy and more cpus.Old sandy is still pretty good and you can oc it to 5Ghz then most likely it will be faster than OC 4790k

 

[DrMrLordX]

Thanks for adding that pic. The 860k does alright for a budget CPU. The minimum frames stay above 30 which is just fine for the price. In fact, the tiny delta between minimum and average frames on the 860k (and most other CPUs) is quite interesting.

 

[frozentundra123456]

Looks like they are getting gpu limited among the top six or so processors.

Edit. Would be nice to see a frame time graph for the 3258. The data looks really strange. I know it stutters, but it cant dip to 11 very often and still get an average of 40 (the overclocked result).

As for Sandy vs Haswell, I dont think a lot of SB will get to 5.0 day to day use that easily, and as I said, the upper end is looking gpu bound with only a 970 anyway, so I dont think this test tells you much.

The game.gpu test, with SLI 980s showed stock 4670k vs 2500k about 25% faster. So even if you generously give the 2500k +10% overclock, it would still be quite a bit slower.

 

[BigDaveX]

It'd be interesting to see whether Nehalem can still deliver playable framerates in this game. Sandy Bridge still holds up very well, but I'm guessing Nehalem will probably be more around the level of the faster Piledrivers (Gulftown might be faster, seeing as it's a hex-core).

 

[Burpo]

Of course it can. I'm using an overclocked Westmere EP Xeon X5650 hexcore & it performs on a par with i7-3770K in single thread and blows it away on multi-thread.

 

[Sweepr]

It'd be interesting to see whether Nehalem can still deliver playable framerates in this game. Sandy Bridge still holds up very well, but I'm guessing Nehalem will probably be more around the level of the faster Piledrivers (Gulftown might be faster, seeing as it's a hex-core).

Looking at PCGameHardware I think overclocked Nehalem should be able to ouperform FX.
After 7 years, a 4GHz Core i7 920 still packs a punch. Great chip.

 

[TurtleCrusher]

Looking at PCGameHardware I think overclocked Nehalem should be able to ouperform FX.
After 7 years, a 4GHz Core i7 920 still packs a punch. Great chip.

A CPU with 325w+ power consumption. Whew! Great chip indeed.

 

[Burpo]

Xeon X5650 @ 4.4Ghz <180 watts

 

[Atreidin]

I don't think anyone reasonable expected an old design to suddenly start topping charts. Maybe age gracefully with the right software, at least, meaning at least staying moderately usable for a while.

But those Pentium results, wow. Looks like garbage. Nobody could accuse that CPU of aging gracefully for budget gaming systems.

 

[Sweepr]

A CPU with 325w+ power consumption. Whew! Great chip indeed.

Please share the source of those power consumption results with us, I'm sure you have quite a lot of experience with power hogs.

4GHz Core i7 920 should perform about the same as stock Core i7 2600K. Looking at OCed Phenom II at the bottom (sub 30 FPS framerate, ouch) and current FX chips that were supposed to trash Haswell Core i5 @ GTA V - Bloomfield/Gulftown are aging quite well.

 

[seitur]

hmmmm

could be patched in.

my understanding was DX12 is basically a software change in a more efficient rendering pipeline, and basically all the DX11.1 cards are ready to go with DX12.

No DX12 is completly diffrent from DX11. GTA 5 on DX12 is not gonna happen.

 

[Blitzvogel]

Please share the source of those power consumption results with us, I'm sure you have quite a lot of experience with power hogs.

4GHz Core i7 920 should perform about the same as stock Core i7 2600K. Looking at OCed Phenom II at the bottom (sub 30 FPS framerate, ouch) and current FX chips that were supposed to trash Haswell Core i5 @ GTA V - Bloomfield/Gulftown are aging quite well.

Trash Haswell based on what? A bunch of narrow integer cores with split FPUs like on FX simply can't hold a candle to proper wide cores with big SIMD like in the last few Intel generations. AMD's only advantage with FX is that PS4 and Xbone developers can essentially take the threads they got running on those consoles and apply them to FX without little consequence. The thing is that a single Intel core can handle more than an AMD module can per clock cycle, especially when you factor in all the "command and control" latency that likely happens when using moar cores.

 

[Magic Carpet]

But those Pentium results, wow.

Some people are even experiencing graphical anomalies. Hopefully, this is patchable. You can tell the gameplay isn't fluid at all, although that Pentium was running stock. You definitely want 4 cores for this game.

 

[Phynaz]

Heat isn't an issue with a cheap aftermarket cooler (freezer pro 7). Even with prime on all cores, 4.3Ghz at 1.4V it doesn't crack 60C. Gaming Starcraft 2 I rarely see it crack 42C and in NBA 2K15 (which is more demanding than this game) it doesn't ever crack ~54C. During cut scenes the CPU usage spikes to ~90% at times.

Temperature <> heat output. You should know this.

 

[soccerballtux]

Interesting that FX-4310 @3.8 GHz beats X4 965 @ 4.0 GHz. That should not happen, I would think?

yeah. that is pretty strange. In fact there's no reason for it, and now I'm questioning if they made CPU-specific customizations/tweaks.