**Official 3DMarkVantage thread**

[Nickolp1974]

if only i had a 3930 for this

 

[Geeman1979]

P56323 Score
Graphics Score
59180
CPU Score
49198

Cpu i7-3960X @ 4.8GHz
1x GTX 690

http://www.3dmark.com/3dmv/4660369

 

[Gregster]

Scores updated

 

[Geeman1979]

Down on just over 2000 points on Kaap's overhall P'score! Dont know why the difference is so great.

 

[Nickolp1974]

Scores updated

thx gregster got a 3930 comming soon so see what i can do then

 

[Kaapstad]

Down on just over 2000 points on Kaap's overhall P'score! Dont know why the difference is so great.

In this bench CPU speed and fast tight memory really distort the final score.

On the standard settings it is not even worth me running my Titans as the CPU holds them back too much and the results are dreadful, on the Extreme preset things are a little different as the Titans can get to work.

http://www.3dmark.com/hall-of-fame-2/3dmark+vantage+3dmark+score+extreme+preset/version+1.1

 

[Geeman1979]

In this bench CPU speed and fast tight memory really distort the final score.

On the standard settings it is not even worth me running my Titans as the CPU holds them back too much and the results are dreadful, on the Extreme preset things are a little different as the Titans can get to work.

http://www.3dmark.com/hall-of-fame-2/3dmark+vantage+3dmark+score+extreme+preset/version+1.1

Yeah memory tuning is something im not up to speed with. I've noticed your graphics score is a lot higher also.

 

[Geeman1979]

What speed was your memory running at Kaap?

 

[whyscotty]

thx gregster got a 3930 comming soon so see what i can do then

Gratz on your 3930k, I can see some serious benching scores coming soon

What you doing with your 3820?

 

[Kaapstad]

Yeah memory tuning is something im not up to speed with. I've noticed your graphics score is a lot higher also.

It is the CPU/memory pulling it up more than the GPUs doing anything extra.

Here is a couple of results to show how much the CPU can hold you back

Below is the X preset

http://www.3dmark.com/3dmv/4636571

Score X64366

NVIDIA GeForce GTX Titan(4x)

Intel Core i7-3930K Processor @5.0

Graphics Score 65323

CPU Score 50351



This is the P preset used here

http://www.3dmark.com/3dmv/4661836

Score P65998

NVIDIA GeForce GTX Titan(4x)

Intel Core i7-3930K Processor @5.1

Graphics Score 72905

CPU Score 51393


As you can see the overall scores are not much different X64366 vs P65998 and would have been even closer if I had run the CPU @5.1 on the X preset .

Also there is not a huge difference in the graphics scores switching presets.

This bench is all about how high you can clock your CPU/memory sadly

@Gregster if you want to you can use the P preset score for the scoreboard

 

[Geeman1979]

Ah! I'll work a little on improving the Cpu speed then. Are you clocking on just the Multiplier or using a combination of Multiplier and bclk? Drivers used for this run were the 306.23 one's.

 

[Kaapstad]

Ah! I'll work a little on improving the Cpu speed then. Are you clocking on just the Multiplier or using a combination of Multiplier and bclk? Drivers used for this run were the 306.23 one's.

For this bench I normally use the 100 strap multiplier.

You can score a tiny bit more on graphics using the 125 strap and raising the bclk a bit to around 129 but most of the time I find it is not worth it as it makes the system unstable.

 

[Gregster]

For this bench I normally use the 100 strap multiplier.

You can score a tiny bit more on graphics using the 125 strap and raising the bclk a bit to around 129 but most of the time I find it is not worth it as it makes the system unstable.

I found the 125 strap and bclk was the sweet spot for me. I wasn't stable enough on 129 and 100 just didn't net enough points on the GPU.

 

[Geeman1979]

For this bench I normally use the 100 strap multiplier.

You can score a tiny bit more on graphics using the 125 strap and raising the bclk a bit to around 129 but most of the time I find it is not worth it as it makes the system unstable.

Same here! I need to work on ram tuning aswell. Ram speed currently running at 2400MHz timings at 11,12,11,32 1T.

Which benifits benchmarks the most, higher speeds with loser timings or lower speeds with tighter timings?

 

[Kaapstad]

I found the 125 strap and bclk was the sweet spot for me. I wasn't stable enough on 129 and 100 just didn't net enough points on the GPU.

The problem with raising the bclk is it effects a lot of stuff line the PCI-E slots and puts more strain on the IMC.

8 Pack showed me something interesting the other day, when running 4 cards benching, use PCI-E 2.0 not 3.0

This sounds crazy but it works, it makes the CPU more stable @high overclocks as the CPU has less work/stress with PCI-E 2.0, I find even though the cards are running less efficiently they still get more done due to the CPU running more stable.

 

[Kaapstad]

Same here! I need to work on ram tuning aswell. Ram speed currently running at 2400MHz timings at 11,12,11,32 1T.

Which benifits benchmarks the most, higher speeds with loser timings or lower speeds with tighter timings?

I am not sure, I find they both help. If I had to choose I would say getting the timings tight is slightly more important then comes memory speed.

I run my memory @2400mhz 9-11-11-25 1T 96 24/7 but when I get in from work tomorrow I will try a few runs on the 3dmark11 physics test with different speeds and timings. If you have any you want me to try please let me know. I will do a run at the timings you are using but if you have any more please let know and also what CPU speed you want.

 

[Gregster]

The problem with raising the bclk is it effects a lot of stuff line the PCI-E slots and puts more strain on the IMC.

8 Pack showed me something interesting the other day, when running 4 cards benching, use PCI-E 2.0 not 3.0

This sounds crazy but it works, it makes the CPU more stable @high overclocks as the CPU has less work/stress with PCI-E 2.0, I find even though the cards are running less efficiently they still get more done due to the CPU running more stable.

It makes perfect sense. PCI-E 3.0 will deffo put more stress on the cards because they are working harder (or should be). In a 4 way configuration, I could see 2.0 coming out better but in a single card, 3.0 should be better.

 

[Geeman1979]

I am not sure, I find they both help. If I had to choose I would say getting the timings tight is slightly more important then comes memory speed.

I run my memory @2400mhz 9-11-11-25 1T 96 24/7 but when I get in from work tomorrow I will try a few runs on the 3dmark11 physics test with different speeds and timings. If you have any you want me to try please let me know. I will do a run at the timings you are using but if you have any more please let know and also what CPU speed you want.

Nice one

Just ran 3dmark11 at 4.9GHz with the ram settings I mensioned and got through it. Allthough I dont think my rams going to get down to the timings your using. Could do with your overclock settings ideally for 4.9GHz to compare. By the looks of it, our Cpu's look pretty equally matched. I've not even attempted pushing past 4.9 as of yet. What revision of Bios are you guys using btw?

 

[Tonester0011]

I'll play with my 3930k later tonight when it gets a bit quieter.
Only got 4.5Ghz and 2133 9-11-11-28 1T 113...whatever my ram defaulted at.

What kind of voltage are you guys needing for 5Ghz btw? and what kind of LLC.
I'm using a pretty shocking Asrock Extreme9 board (getting RIVE end of month)

 

[Gregster]

I run 1.495V for a stable 5Ghz (1.34V for 24/7 4.625 clocks). I have naff memory that won't budge past 2333Mhz with slack timings but it does net higher scores than at tighter 2133Mhz stock settings.