UPDATED, Part 4 of Benches out! E8400 vs Core i7 with my 5970,5870 and 5850

[Apocalypse23]

Hey, You realize that if you run tripple cards on that board, you will be running PCIe 2.0 at 8x. This would be identical to the bandwidth provided by PCIe 1.0 16x (Gen 1).

"(Note 2) The PCIEX8_1 and PCIEX8_2 slots share bandwidth with the PCIEX16_1 and PCIEX16_2 slots espectively. When PCIEX8_1 is populated with an expansion card, the PCIEX16_1 slot will operate at up to x8 mode; when PCIEX8_2 is populated with an expansion card, the PC IEX16_2 slot will operate at up to x8 mode."

Also, you might as well do 8AA

Yeah, the slots that I will be using are PCIEX16_1, PCIEX16_2 AND PCIEX8_2 for tri-fire, therefore PCIEX16_2 will be running at 8x. So all 3 slots would be operating at 16x/8x/8x. I don't expect much of a performance hit with this however. ;D

My board:

Also, I'll try 8xAA as well.

 

[RussianSensation]

Ya there wont' be a serious hit from 8x. I was just implying that the improved performance will likely come from using a Core i7 @ 4.0ghz as opposed to an E8400 @ 3.9ghz. With 5970 + 5870, we could start to see a big difference from the faster i7.

 

[Sho'Nuff]

Ya there wont' be a serious hit from 8x. I was just implying that the improved performance will likely come from using a Core i7 @ 4.0ghz as opposed to an E8400 @ 3.9ghz. With 5970 + 5870, we could start to see a big difference from the faster i7.

There is a huge difference in performance between a 5970 paired with an e8400 and one paired with an i7. As I posted earlier, my 5970 (which is paired with an i7 860 @~3.8 ghz) is posting numbers over 40% higher than apocalypse's e8400. Why tri-fire, he might still be cpu bottlenecked, but we'll see. Looking forward to seeing the results.

 

[RussianSensation]

Why tri-fire, he might still be cpu bottlenecked, but we'll see. Looking forward to seeing the results.

He bought an i7 though

Don't tell the people that Core i7 3.8ghz provides a tangible performance benefit. It was tried in this thread to no avail: http://forums.anandtech.com/showthread.php?t=2065258

Apparently, C2D/Q and Phenoms are all equally fast enough for 5870 CF.

 

[Apocalypse23]

Here's another update, the i7 920 and OCZ 6GB DDr3 RAM is somewhere held up by Canada Customs, and has been so for about two weeks almost as per it's tracking number by USPS...it's probably going to take a while to get here.

Due to being extremely bored and annoyed by waiting I couldn't resist a 1 day deal of the i7 930 selling for $267.77 CAD here.

I also ordered better quality RAM for $182 CAD : Crucial Ballistix 6GB (3 x 2GB) 240-Pin DDR3 SDRAM DDR3 1600 (PC3 12800)

Now hopefully the stuff will get here by the end of the coming week...i7 930 ftw..I'll just have to sell the i7 920 and cheap OCZ memory whenever they get here...

 

[Makaveli]

why didn't you just order from Ncix which is in your province?

 

[Apocalypse23]

why didn't you just order from Ncix which is in your province?

I got a deal on the i7920 and OCZ ram from AT Sales forums, it cost me a lot less, but the shipment is taking too long. I saved on the PST on my i7 930 on a 2 day only killer deal from PC Cyber, and also, the Crucial RAM is not in stock currently at the nearest NCIX outlet, so it had to be Newegg.

 

[Apocalypse23]

Another Update: The i7920 and OCZ DDR3 RAM shipment seems to be lost as the tracking information shows it arrived in Vancouver Customs on the 7th, and USPS and Canada Post have launched an inquiry on them.

The i7930 should arrive by Friday, but the Crucial Ballistic DDR3 6GB sticks from Newegg.ca are arriving here on the 29th. So I'm hoping that I get the "lost" shipment before the 29th or else my i7 build will take place on the 29th.

Sigh, this has been a long process

 

[MrK6]

Damn that sucks. Best of luck in it getting through the red tape ASAP. I know I hate waiting for parts to start a new build.

 

[Apocalypse23]

Damn that sucks. Best of luck in it getting through the red tape ASAP. I know I hate waiting for parts to start a new build.

Thanks, 2 more days to go until I get the RAM, got the i7 930 today. Also just got the i7 920 and the OCZ RAM today. I'll build the i7 920 and test it out first, and then install the i7 930 later as that will be the one I will be keeping.

 

[dopee123]

suscribed

 

[Apocalypse23]

Conducting benchmarks at stock clock atm, I still gota OC this thing.

 

[Apocalypse23]

Edit: E8400@3.9Ghz vs i7 930@stock with turbo off conducted on Asus 5870:

Comments: Overall the i7's stock performance felt way smoother while playing, you could sense that although the fps may be low, the game felt crisp as ever, the Quads are really helping. The FPS alone may not give you the overall picture. Further testing at 4.3Ghz speed and CF benches to come this weekend.

 

[Lonyo]

Could you do the E8400 at stock as well? (pretty please)

 

[Genx87]

Oh wow it is possible to run an E8400 at a speed that will compete with the i7. Might be worth just upgrading my GPU instead of a whole machine then. E8400s are pretty notorious for overclocking.

But a stock speed comparison would also be awesome.

 

[Termie]

Wow, surprisingly big improvement here, even at the stock clocks of the 930. Very impressive. Didn't realize that even Crysis would take advantage of quad cores more than clock speed. By the way, I think you might want to change the scale on the Just Cause benchmark - should probably start at zero like the others.

 

[Apocalypse23]

Could you do the E8400 at stock as well? (pretty please)

 

[swilli89]

Wow that E8400 at stock speeds really gets hammered in the Min frames category in everything tested here.

Darn it, I was trying to convince myself to hold on to the e6600 @ 3.0 for a bit longer..

 

[Lonyo]

Wow that E8400 at stock speeds really gets hammered in the Min frames category in everything tested here.

Darn it, I was trying to convince myself to hold on to the e6600 @ 3.0 for a bit longer..

And what does it tell you?

It tells me that sites which test cards with a 4GHz Core i7 don't give useful recommendations to people using something like a 3GHz Core 2 as their CPU.
Sure, a GTX480 might have better minimums than an HD5870 when you are GPU limited, but you have to be GPU limited.

Someone on a slower CPU would be better off either saving their money, or making sure they overclock the poop out of their CPU, because it's not going to do justice to a GTX480.

This is why review sites need to do more testing, rather than seeing which company makes the fastest GPU when the GPU is the only limiting factor. It's not useful for the more typical consumer, only for those who overclock heavily.


And thanks Apocalypse23 for the tests! Some useful information for once, rather than what the review sites show.
Now we need to get someone with an AMD system to do the same. And some of the lower end Core ix machines, like the i3 and i5s.

 

[Apocalypse23]

Crucial RAM arrived yesterday, reapplying thermal paste as this chip is running hot when trying to hit 4.2/4.3ghz..

 

[cusideabelincoln]

I love that case.

 

[tviceman]

And what does it tell you?

It tells me that sites which test cards with a 4GHz Core i7 don't give useful recommendations to people using something like a 3GHz Core 2 as their CPU.
Sure, a GTX480 might have better minimums than an HD5870 when you are GPU limited, but you have to be GPU limited.

Someone on a slower CPU would be better off either saving their money, or making sure they overclock the poop out of their CPU, because it's not going to do justice to a GTX480.

This is why review sites need to do more testing, rather than seeing which company makes the fastest GPU when the GPU is the only limiting factor. It's not useful for the more typical consumer, only for those who overclock heavily.

I agree and disagree with this. If I see a review or showdown of the best GPU's available, I think everything should be done to remove potential bottlenecks. I think it's also great - as you point out - to have a frame of reference to see how the same GPU's perform on lesser, stock clocked processors, but I do think buying a high end GPU for a CPU-limited system can be, in a sense, a future investment for some.

 

[Genx87]

Thanks a lot for the updated stock benches. So it appears at stock there is quite a jump in performance with the newest GPU's between an e8400 and i7. Just need to determine if I want to overclock or not.

 

[Lonyo]

I agree and disagree with this. If I see a review or showdown of the best GPU's available, I think everything should be done to remove potential bottlenecks. I think it's also great - as you point out - to have a frame of reference to see how the same GPU's perform on lesser, stock clocked processors, but I do think buying a high end GPU for a CPU-limited system can be, in a sense, a future investment for some.

I totally agree that review sites should showcase them in their best light, GPU limited, try and remove every bottleneck they can.

But for most consumers, that's not particularly informative. You pretty much need to do both.
Most people don't have SSDs, most people don't have 4GHz i7's, most people have regular computers.

If there are two cards which seem close in performance when tested as review sites test them, that might not play out in a real world situation where you have a slow hard drive, a slower processor etc. That's what people need to know as well, and that is especially true when you move away from the super high end.
Thinking someone who's going to purchase SLI GTX480s or an HD5970 would have an i7 and maybe overclock it and maybe have an SSD, that's reasonable.
To assume someone buying an HD5850 or GTX would have the same? Not so much. Those are more reasonably priced cards, and people with normal systems might want to buy them, and if someone says to get one over the other because it seems to perform better (e.g. GTX470 over HD5850) but for a realistic system it doesn't perform any better, the person would be wasting money.
Same could go for GTX470 vs HD5870. If you have $400, you might think to get the "faster" HD5870, but it might not actually be any faster on a real computer, they might perform equally well, so you'd be throwing away $50. Or maybe you would be throwing away more because even an HD5850 would struggle, and you should spend $150 on a new CPU instead.

Basically most reviews are bunk for regular people in terms of being good buying guides. They are only really good for people who are at the high end with top everything, since they don't take into consideration potential things like people running lesser processors.

 

[Apocalypse23]

Thanks for the warm wishes guys, I'll be starting the CF benches tomorrow. I managed to OC this thing but it took me a long time to find stable settings. Prime 95 was stable but BFBC2 would crash like crazy at settings, which forced me to increase the Vcore and thus I was priming at 97C max at one point for one core... Anyhow the setting seems stable. I'm still fine tuning and cutting down on any extra voltages atm to lower the overall temps, the Vcore however will stay the same.

Test System:
Intel Core i7 930
Gigabyte X58A-UD3R
Crucial Ballistix DDR3 1600 2*3 6GB 8-8-8-24 2T RAM
Scythe Mugen 2 Rev B Heatsink
Corsair TX950 950W PSU
Asus 5870/Asus 5970/Asus 5850
Coolermaster Storm Sniper Black Edition Case
Asus 24x DVD ROM



New benches: