Radeon HD 7970: Bulldozer vs. Sandy Bridge vs. Nehalem

[qubit]

Originally Posted by W1zzard

all benchmarks are dx11. except for hard reset and star craft II

Then it will definitely be interesting to see those benchies after the scheduling/threading patch comes out.

[FreedomEclipse]

Originally Posted by qubit

Then it will definitely be interesting to see those benchies after the scheduling/threading patch comes out.

I wont hold my breath, as many have said - you cant polish a turd.

Its like back in the day when game devs used to release dualcore optimisation patches for their games, I cant remember any of them patches really giving a huge boost in frame rate

Programs are written for multiple cores these days. All the multicore processors didnt need a patch a side from AMDs dualcore optimiser.

what we have here is a multicore patch being written for a multicore program because the CPU aint multicore enough and handles like driving a car through pig shit up to your knees. Intel didnt need this patch and im sure the multicore AMD cpus didnt need it either to this one.

In a small sense, I see this as cheating. their horse lost the race so they bring it back to the stalls and pump it full of steriods,

the patch wont bring the results that people are hoping for and any good results will be heavily skewed making the results either hard to believe or totally untrustworthy

[eidairaman1]

zzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzz, broken record anyone?

[qubit]

Originally Posted by FreedomEclipse

I wont hold my breath, as many have said - you cant polish a turd.

Its like back in the day when game devs used to release dualcore optimisation patches for their games, I cant remember any of them patches really giving a huge boost in frame rate

Programs are written for multiple cores these days. All the multicore processors didnt need a patch a side from AMDs dualcore optimiser.

what we have here is a multicore patch being written for a multicore program because the CPU aint multicore enough and handles like driving a car through pig shit up to your knees. Intel didnt need this patch and im sure the multicore AMD cpus didnt need it either to this one.

In a small sense, I see this as cheating. their horse lost the race so they bring it back to the stalls and pump it full of steriods,

the patch wont bring the results that people are hoping for and any good results will be heavily skewed making the results either hard to believe or totally untrustworthy

Yes indeed, I agree that BD shouldn't need this and the other points you make and I'm not holding my breath at all. It's just interesting to see the results, whatever they show - and you can bet the threads on this will be long and ranty strongly worded.

[Mussels]

Originally Posted by W1zzard

all benchmarks are dx11. except for hard reset and star craft II

i spose clear majority is good enough :P and probably why BD performed better than in some other reviews where people hated on it.

[EarthDog]

I havent read the thread so my apologies is this was asked/answered before.

First, Excellent review!

I would have loved to have seen a 2600k and 1100t?
Was the BD patch for W7 used? I cant imagine it making much of a difference but, its available, so....
Ram speed looks underclocked? Not that it makes a difference in most cases but....1333Mhz on AMD BD and especially 1066 on SB are awfully low.

Again sorry if I missed the info if it was listed anywhere else.

[OOZMAN]

Originally Posted by EarthDog

I havent read the thread so my apologies is this was asked/answered before.

First, Excellent review!

I would have loved to have seen a 2600k and 1100t?
Was the BD patch for W7 used? I cant imagine it making much of a difference but, its available, so....
Ram speed looks underclocked? Not that it makes a difference in most cases but....1333Mhz on AMD BD and especially 1066 on SB are awfully low.

Again sorry if I missed the info if it was listed anywhere else.

Meh... 2600k performs the same as the 2500k in games... 1100t on the other hand, would've been interesting.

[Lost Hatter]

Originally Posted by Dj-ElectriC

2500K, Chuck norris use this CPU.
thanks for the megabench w1zz, i know how long and frustrating are all those tests

Chuck Norris Runs a 486-dx2-66. But gets the speeds of a i7-3960x

[ensabrenoir]

First it was wait till bulldozer, now its wait till the patch, then itll be wait till piledriver...and on and on. Bulldozer only fails when its trying to beat intel. It is what it is. If u have one..... its yours. Just enjoy the darn thing. Careful for what u ask for. Someone is gonna make a patch that,pushes bd to the ultimate and your gonna have a gigabite experience.

[Lost Hatter]

Actually Chuck doesnt need a CPU at all. His computer runs without 1. out of fear for its life.

[Yellow&Nerdy?]

Bulldozer has highest default clock, but still gets beat. This really shows, that the 2500K is the gamer's choice for CPU: it's faster, cheaper and consumes less power than Bulldozer.

[nINJAkECIL]

Looking at the Skyrim result (all cpu across all resolution), I do wonder whether this 7970 is too much overkill even for 2500K. At the lowest resolution up until max resolution, there's barely any difference. Is this a sign that this game eats cpu for breakfast?

[ensabrenoir]

Originally Posted by Yellow&Nerdy?

Bulldozer has highest default clock, but still gets beat. This really shows, that the 2500K is the gamer's choice for CPU: it's faster, cheaper and consumes less power than Bulldozer.

Careful partner .....the truth aint welcome in these here, parts....amd has a patch for that kinda thinking.

[Wyverex]

Originally Posted by qubit

Excellent review.

Unfortunately, Bulldozer is what it is and it doesn't matter what system you put it in, you get about the same results.

I look forward to an updated version of this review once the Windows threading patch comes out.

Ditto!

[bear jesus]

I find it kind of funny that to me (gaming at over 2560x1600) the difference between sandy bridge and bulldozer is about 2fps down to 0.1fps, who here can see a 0.1fps difference? or even 2fps difference?

I admit for low res usage then of course sandy bridge pulls way ahead but who would be using sandy bridge and a 7970 at 1280x1024?

It seams that for the most part playing games at high settings and resolution makes them so GPU limited it does not really matter what CPU is being used so in that case the people who bought bulldozer are getting the same gaming performance as people who bought sandy bridge.

[W1zzard]

Originally Posted by bear jesus

so GPU limited it does not really matter what CPU is being used

well, the cpu just needs to be fast enough to make the rendering gpu limited. those are basically the two options. either cpu limited or gpu limited.

[bear jesus]

Originally Posted by W1zzard

well, the cpu just needs to be fast enough to make the rendering gpu limited. those are basically the two options. either cpu limited or gpu limited.

It seams that most modern CPUs are so powerful that when it comes to gaming the odds of it being CPU limited are pretty low and using a higher resolution just pushes the GPU more, so what i am getting from this is that the testing might have shown me more improvement for my situation if it was being done with an 8970 or 9970

Really what this shows me is that all modern high end CPUs are damn powerful and that as far as high resolution gaming goes i should concentrate way more on the GPU than the CPU.

[Hayder_Master]

Great idea, awesome review w1z.

LOl, that's mean if u try SB I7 it's put any CPU away in all tests.

Cograts intel.

[mlee49]

I'd love to see more reviews like this. An awesome move from the standard product reviews.

[theoneandonlymrk]

firstly good review wizzard, must have taken some time that one but worth it in the end imho, nice to see the BD bashers arent bored yet too , come on peeps BD aint all that for sure but at higher gameing res's it isnt as bad as some of your are portraying, 10% down on 1 2500K isnt that bad to me and thats an average, again as more multi threaded games come out the cores will start to count more

and to freedom intel might not need a patch per say but recent BF3 hyperthreading jitters and the crappy chipset mobo recall recently inform me that the intel worlds not perfect either and only a crack smuggler or meph fiend would think the sun shined out of either companys rear end anyway, sort your bias peeps, i vote you jump on the consumers side as opposed to either co's that way you can slate them both from time to time like me

[cdawall]

Originally Posted by FreedomEclipse

I wont hold my breath, as many have said - you cant polish a turd.

Its like back in the day when game devs used to release dualcore optimisation patches for their games, I cant remember any of them patches really giving a huge boost in frame rate

Programs are written for multiple cores these days. All the multicore processors didnt need a patch a side from AMDs dualcore optimiser.

what we have here is a multicore patch being written for a multicore program because the CPU aint multicore enough and handles like driving a car through pig shit up to your knees. Intel didnt need this patch and im sure the multicore AMD cpus didnt need it either to this one.

In a small sense, I see this as cheating. their horse lost the race so they bring it back to the stalls and pump it full of steriods,

the patch wont bring the results that people are hoping for and any good results will be heavily skewed making the results either hard to believe or totally untrustworthy

So if the patch works its cheating? But the dual core optimisor was ok? Wtf?

[trt740]

Originally Posted by Andrei23

Nice review. but bulldozer is still a fail for gaming.

Hardly, it appears to me that the cpu makes very little difference at most gaming resolutions, and it is more about GPU than cpu. 6-10 frames is not that big a deal and anything over 40 Fps sustained make zero difference. All these cpus are fast as heck buy the cheapest cpu and the most expensive gpu you can afford.

[PopcornMachine]

Originally Posted by trt740

Hardly, it appears to me that the cpu makes very little difference at most gaming resolutions, and it is more about GPU than cpu. 6-10 frames is not that big a deal and anything over 40 Fps sustained make zero difference. All these cpus are fast as heck buy the cheapest cpu and the most expensive gpu you can afford.

Yes many of the tests were within margin of error, or ties.

But when there was separation, it was the 2500k doing best.

And while these may not all have been noticeable differences, I'm happy I have the more powerful and cheaper and more power efficient cpu.

The main thing to realize is that not only does the bulldozer not beat the I5 2500K, it pretty much is matched by the I7 920. And it would have been nice to see a 6-core phenom II in here. Bet it would have given it all it could handle too.

Not what AMD was hoping for from it's new flagship CPU.

[qubit]

Originally Posted by bear jesus

I find it kind of funny that to me (gaming at over 2560x1600) the difference...

You game at a resolution I can only dream of.

I hate you.

End.

j/k

Oh and 1280x1024? I play at 640x480 sometimes just for that 'pixellated experience'.

[FreedomEclipse]

Originally Posted by cdawall

So if the patch works its cheating? But the dual core optimisor was ok? Wtf?

Of course.... Dual core optimiser was more of a patch that helped 'sync' some programs that had some timing issues when it came to dual core.

A good example of this would be CoD:UO - the first map in the SP stage where your in the forrest. sometimes things wont spawn or scripts that are ment to happen at a certain time dont happen at all.

It didnt really boost performance of Windows per se but when it came to programs that need to bypass Windows API for timing thus fixing some of the problems some games were having.

plus even WITHOUT the dual core optimiser patch installed, it hardly made a difference to benchmarks at all in my experience. you could live without it and never know it existed.

People are looking at this patch from M$ like its some holy grail of patches and its gonna bring balance to the force when its not going to do shit all. If M$ are truely capable of making a CPU shine just by patching it then why the hell would we need such powerful processors??? We could buy any cheap ass low-end multicored CPU and just have M$ write a patch that optimises the hell out of it.

With that being said... you could put Michael Schumacher in a banged up Ford Cortina and he will drive like a boss and put on a world class performance, but at the same time. he wont win any races. because a driver is only as good as his car. and the car totally sucks at this juncture of all junctures.