Tech and Talk > Gaming > PC gaming Tech > New benchmarks for th 8350 and 3570k/3770k. Weird results.

New benchmarks for th 8350 and 3570k/3770k. Weird results.

Can you summarize for those of us behind the great wall?

Basically in steaming(xsplit) the 8350 won, and for a lot of other games(most he did, including metro 2033) the 8350 won by a decent bit, nearly 10fps or so. Same gpu was used for the benches.

reply to me1212

reply to me1212
I got a chance to watch this today from home, and I am surprised, as he is contradicting every other site out there.

But he's using a 7870 as the GPU? He should have been using a top-end GPU like the 680 or 7970 so he can determine how much the CPU is limiting the rest of the rig. I'm not sure why he used a mid-range GPU.

But testing Black Mesa Source? When you are trying to make differences between hundreds of FPS, the difference is going to be coding or drivers implementation, not CPU prowess.

I would be interested in an Anandtech or Tom's Hardware independent peer review of his findings so they have a chance to update their own or refute his findings.

This is not enough for me to changing to recommending the FX-8350--I would bet this has more to do with using the mid-range GPU than the CPUs.

What I really would be interested in is an OC matchup between the FX-8350 and the i5-3570K on the same cooler, like an H100i.

Also, the biblical doubting reference is Thomas, not Judas. lol.
--
Those who would give up Essential Liberty to purchase a little Temporary Safety, deserve neither Liberty nor Safety.

The 7078 puzzled me too, I know he has a 680 in his office he could have used, but he did give away the 7970. Still even a mid range cpu shouldn't make a difference like that. Not that I'm an intel fanboy, heck I use amd's more core cpus on my code compiling box.

I'm guessing he did black mesa source because its free.

It's still at odds with what everyone else has gotten in their tests. There are three possibilities:

1) His testing is "better" in terms of consistency and methodology than all the other review sites out there.
2) There is a conspiracy by everyone else to push Intel CPUs even though they are subpar.
3) His testing results are flawed.

I am going with #3 if they can't be verified by anyone else out there.
--
Those who would give up Essential Liberty to purchase a little Temporary Safety, deserve neither Liberty nor Safety.

I think it may be 3 aswell, based on what I've read on his site he didn't do a reinstall when he switched cpu+mobo and he did the amd first. Just switched the hardware and installed the drivers.

lolwut? He just stuck the SSD into a completely new system?

Ether that or swapped out the cpu and mobo, possibly ram too. Ether way, just why would one do that?

He's cutting corners, I guess.

I could see pulling out the GPU, as that would make no difference, but he shouldn't be moving the SSD between systems--I'm surprised the OS install even worked and didn't detect new hardware and shut down and ask for a new license key.

If he is going to cut corners, he has no business making claims or publishing benchmarks. The goal is in testing is to remove as many variables as possible. That usually means a fully-functioning, fully-patched system running the latest stable version of drivers.

Shifting hardware around like that only increases the entropy he's introducing into his systems.
--
Those who would give up Essential Liberty to purchase a little Temporary Safety, deserve neither Liberty nor Safety.

If he wasn't using a legit copy of windows 7 it may not have asked him for new key.

That's speculation at this point. I'll just mark these results as "suspect" and move along.

I'm not trying to say he was, I'm just saying that some cracked copies don't ask.

reply to me1212
If you install Windows 7 without a key you can evaluate it for up to 120 days without activation. If you don't activate it won't prompt you for a key after a hardware change.

Windows 8 install media wants to force activation directly after OS install and requires a key for the install. I hear you can get around this and install to evaluate as well but it takes some work,...

reply to me1212
For what its worth, heres a review done in direct response to the first one I posted. I'm at college right now so I haven't had time to watch it all yet.

»www.youtube.com/watch?feature=pl···YV8Djt7k

I understand the criticisms, in the chain and the second vid (though I disapprove of the attitude and name calling in the second vid), and agree that the original test is not rigorous and really doesn't say which is the "better" cpu. Viewed in light of the criticism, I think the the first video only shows if actual game performance is your main concern, and you can't afford a high-end GPU (and don't plan to upgrade the component parts), you can save money by choosing the 8350. This strikes me as intuitive -- if the system is GPU bound, having a "better" gaming CPU will not provide much better gaming performance, if any.

I think what would be more interesting is to test which GPUs you would need for the CPU to make a difference. That would be much more helpful for a gamer trying to build a system to get the best bang for the buck. You could then figure out "I can spend the $X difference in MB/CPU on this GPU upgrade" or "I can spend the $Y difference in GPUs on the Intel" and get better performance. Unfortunately, I suspect that nobody is going to spend to money to do that kind of testing.

reply to me1212
So to save money, you go for the AMD, that generates more heat and uses a lot more power?

And is more expensive with less power for gaming?

More expensive? A quick search prices $200 for the FX-8350 v $330 for the 3770k. Last I heard, $200 is less than $330. You can pay for a lot of electricity for $130. Do you have figures for the electricity differential? How many hours of gameplay do you need before you use a $130 differential in electricity? If it's $.10 per hour (and I'm guessing it's a lot less) then that's 1300 hours of gameplay. I'm thinking that's 2 years of gameplay for me. I'm guessing the differential is actually less than $.01 per hour. That's probably 4 or 5 times the life of the system.

More power for gaming? If both systems are limited by the GPU in real gameplay, the extra computing power is wasted. If the GPU is only going to give you 50fps in your game with either chip, how are you better off spending another $130 for the same framerate?

The point is that if your budget means that you're going to be GPU limited either way, maybe the $130 puts you into a better GPU and gets you better performance for the same dollars for the system as a whole with the FX-8350. If that's the case, you're better off putting the money into the GPU instead of the CPU.

Admittedly, you can get the 3570k for about $220 or so; but if history is any guide, the FX-8350 price will fall faster than the 3570k price. $20 may or may not make a difference in the GPU, but $50 probably will. Would you take a 3570k with a GT 610 over an 8350 with a GT 620? Sure would. What about a 3570k/GT610 over 8350/GT640? maybe, maybe not - tell me what the actual framerates are. 3770k/GT610 v 8350/GT650? I'd take the 8350/GT650.

The point is, the intels have more bang; but depending on the circumstances, the FX-8350/GPU combination for the same money may have more bang for the buck. If your budget means that your GPU is not going to be top of the line, and either CPU is going to be limited by the GPU performance, then get the best GPU you can for the difference to maximize your bang for the buck.

So to answer your questions: Yes. If the FX-8350/better GPU combo gets better framerates on my game than the same priced Intel/worse GPU combo, I'll take the FX-8350.