GTX 670 SLI 3DMark 11 score. Little loss due to PCI-E 2.0?

Toggle sidebar Toggle sidebar
The_Golden_Man

The_Golden_Man

Senior member
Apr 7, 2012
816
1
0
I have a feeling I'm loosing a tiny bit on 3DMark11 due to PCI-E 2.0. I'm limited to PCI-E 2.0 due to using a ASUS P8P67 PRO 3.0 and a 2600K. This means I get x8 + x8 in PCI-E 2.0 when in SLI.

Whilst a system using a Z77 motherboard and a Ivybridge would get PCI-E 3.0 x8 + x8, which is double the Bandwith.

My 2600K is overclocked to 4.4GHz, my two ASUS GTX 670 Direct CU II are running stock. One boosts to 1124MHz and the other boosts to 1084MHz.

I'm using the standard 3DMark11 Performance preset with stretched mode.

Would be interesting to see what score others get using a similar setup utilizing PCI-E 3.0

Using 304.79 Beta drivers.

3DMark11 P Score = 14298

Score
P14298 3DMarks

Graphics Score
17137

Physics Score
10032

Combined Score
8914
 
Last edited:
The_Golden_Man

The_Golden_Man

Senior member
Apr 7, 2012
816
1
0
X

Xed

Golden Member
Nov 15, 2003
1,452
0
71
I'd look at overclocking your memory even if you don't want to mess with core clocks.

I scored 234xx for graphics with those settings with 2 lightnings in sli @ 1300/7000
 
The_Golden_Man

The_Golden_Man

Senior member
Apr 7, 2012
816
1
0
Xed said:
I'd look at overclocking your memory even if you don't want to mess with core clocks.

I scored 234xx for graphics with those settings with 2 lightnings in sli @ 1300/7000
Click to expand...

Increased mem to 6.8GHz on both cards.

So got a tiny difference from 6GHz memclock.

3DMark11 P = 14357

Score
P14357 3DMarks

Graphics Score
17241

Physics Score
10027

Combined Score
8938
 
Last edited:
The_Golden_Man

The_Golden_Man

Senior member
Apr 7, 2012
816
1
0
Clearly my Graphics score are somewhat limited by PCI-E 2.0 x8 + x8 lanes VS cmdrdredd's PCI-E 3.0 x8 + x8 setup.
 
cmdrdredd

cmdrdredd

Lifer
Dec 12, 2001
27,052
357
126
Right, I tested this exact thing and saw similar results. Most games though didn't seem to change much.
 
The_Golden_Man

The_Golden_Man

Senior member
Apr 7, 2012
816
1
0
cmdrdredd said:
Right, I tested this exact thing and saw similar results. Most games though didn't seem to change much.
Click to expand...

Good to hear. Anyway, it's clear that PCI-E 2.0 in x8 + x8 is stretched to it's limits, and a little over the top when using GTX 670 SLI.
 
cmdrdredd

cmdrdredd

Lifer
Dec 12, 2001
27,052
357
126
at 8x yeah, because there is no x16/x16 off the CPU. I'd be interested to see a compare with a PLX equipped motherboard for x16/x16 running 2.0 vs 3.0
 
The_Golden_Man

The_Golden_Man

Senior member
Apr 7, 2012
816
1
0
cmdrdredd said:
at 8x yeah, because there is no x16/x16 off the CPU. I'd be interested to see a compare with a PLX equipped motherboard for x16/x16 running 2.0 vs 3.0
Click to expand...

PCI-E 2.0 x16 + x16 will equal to PCI-E 3.0 x8 + x8. However, a PLX chip will add some latency.
 
MrMuppet

MrMuppet

Senior member
Jun 26, 2012
474
0
0
If even a single overclocked GTX 680 can benefit from PCI-E 3.0 x16 over PCI-E 2.0 x16, I think it's safe to assume that dual GTX 670 in SLI can benefit from PCI-E 3.0 x8 over PCI-E 2.0 x8. Your findings in this thread seem to confirm this assumption.

Vega said:
I did some testing on the Maximus V GENE motherboard with a highly overclocked GTX 680 (1320 MHz core), having it's own dedicated 16x 3.0 slot it was 5% faster than when the slot was bumped down to 8x PCI-E 3.0. So even with a single GPU the differences between 16x 2.0 (8x 3.0) and 16x 3.0 can start to surface. Granted 5% isn't huge but it isn't completely insignificant either and is reproducible as many runs were tested.
Click to expand...
http://hardforum.com/showpost.php?p=1038716287&postcount=347

edit: Noticed this thread was specifically about 3DMark 11 and Vega didn't necessarily test that (can't recall).
 
Last edited:
cmdrdredd

cmdrdredd

Lifer
Dec 12, 2001
27,052
357
126
3dmark 11 has a graphics score that reflects directly with the performance difference between 2.0 and 3.0
 
guskline

guskline

Diamond Member
Apr 17, 2006
5,338
476
126
The_Golden_Man said:
I have a feeling I'm loosing a tiny bit on 3DMark11 due to PCI-E 2.0. I'm limited to PCI-E 2.0 due to using a ASUS P8P67 PRO 3.0 and a 2600K. This means I get x8 + x8 in PCI-E 2.0 when in SLI.

Whilst a system using a Z77 motherboard and a Ivybridge would get PCI-E 3.0 x8 + x8, which is double the Bandwith.

My 2600K is overclocked to 4.4GHz, my two ASUS GTX 670 Direct CU II are running stock. One boosts to 1124MHz and the other boosts to 1084MHz.

I'm using the standard 3DMark11 Performance preset with stretched mode.

Would be interesting to see what score others get using a similar setup utilizing PCI-E 3.0

Using 304.79 Beta drivers.

3DMark11 P Score = 14298

Score
P14298 3DMarks

Graphics Score
17137

Physics Score
10032

Combined Score
8914
Click to expand...

Here are my scores for rig 1 below with 2 EVGA GTX 670 FTWs in SLI. I'm running an Asus P8Z68-V Pro/Gen3 mb with an Intel 2500k OC'd to 4532 (103x44). The rest of the specs in sig. GPUs are stock.

Score

P14129 3Dmarks

Graphic
18562

Physics
83562

Combined
7993

The Golden Man, your scores look fine. Your Physics is likely higher due to a 2600k (hyperthreaded and more ram). We all want the "fastest" scores but tell me these aren't fast HOLY SMOKES these cards are rocket ships. BTW 2 670s in SLI are just what the Dr. ordered for playing BF3 on 3 monitors at 5760 x 1080.
 
Last edited:
cmdrdredd

cmdrdredd

Lifer
Dec 12, 2001
27,052
357
126
guskline said:
Here are my scores for rig 1 below with 2 EVGA GTX 670 FTWs in SLI. I'm running an Asus P8Z68-V Pro/Gen3 mb with an Intel 2500k OC'd to 4532 (103x44). The rest of the specs in sig. GPUs are stock.

Score

P14129 3Dmarks

Graphic
18562

Physics
83562

Combined
7993
Click to expand...

You sure that Physics score is correct?
 
cmdrdredd

cmdrdredd

Lifer
Dec 12, 2001
27,052
357
126
guskline said:
cmdrdredd: How much score did you pick up OCing your GPUs from stock?:)
Click to expand...

I'll run a test at stock and report back.

Your physics score seems really high. I thought you typo'd when you posted.

edit: score 14849, graphics score 18885, physics score 9183, combined score 8855 at stock settings on both GPUs.
 
Last edited:
D

digitaldurandal

Golden Member
Dec 3, 2009
1,828
0
76
Hardocp tests showed a very small difference in frame rate. Obviously if you are building a new rig and can afford it - you do want pci-e 3.0. However you need to be careful because some of the boards have a bug with the 670's that will make them only work at 1.0.

However I really do not think it is anything you need to be stressing over. We are talking about single digit frames here and your system is going to run anything just fine.
 
cmdrdredd

cmdrdredd

Lifer
Dec 12, 2001
27,052
357
126
digitaldurandal said:
Hardocp tests showed a very small difference in frame rate. Obviously if you are building a new rig and can afford it - you do want pci-e 3.0. However you need to be careful because some of the boards have a bug with the 670's that will make them only work at 1.0.

However I really do not think it is anything you need to be stressing over. We are talking about single digit frames here and your system is going to run anything just fine.
Click to expand...

I've only heard about the Gigabyte cards having issues.
 
guskline

guskline

Diamond Member
Apr 17, 2006
5,338
476
126
I loved playing BF3, COD MW3, ROF and Microsoft Flight at 5760 x 1080 and the games were smooth and fast. Incredible power in 2 of these in SLI
 
You must log in or register to reply here.

TRENDING THREADS

COMPANY
RESOURCES
FOLLOW
Top Bottom