[mrbean]

I am sure someone will find a use for this, as I did. I wanted to see what my score would be, if I only used Solo's dualcore cpu, and add it to my system. This is theoretical only, because we all realize, that the more powerful the cpu becomes, the bigger the effect it will have on both SM2 and SM3 scores as well, which will influence (greatly) the final score.

1stly, the Formulas:

3dMark score Calculations:

SM2.0 Score = 120 x 0.5 x (SM2 GT1 fps + SM2 GT2 fps)
HDR/SM3.0 Score = 100 x 0.5 x (SM3 GT1 fps + SM3 GT2 fps)
CPU Score = 2500 x sqrt (CPU1 fps x CPU2 fps)

GS for hardware capable of running all tests = 0.5 x (SM2S + HDRSM3S)
GS for hardware capable of running only SM2.0 graphic tests = 0.75 x SM2S

3dMark Score = 2.5 x 1.0/((1.7/GS + 0.3/CPU Score)/2)
i.e 2.5 weighted harmonic mean of GS and CPU Score

GT1 fps means the average frame rate measured in SM2.0 graphics test1.
CPU fps means the frame rate measured in the cpu tests.

Solo:
X1900XT - 6333
SM2.0 - 2457
SM3.0 - 2656
CPU - 2409

Beanie:
7800GTX SLi - 7064
SM2.0 - 3823
SM3.0 - 3625
CPU - 1194

for Solo:
GS for hardware capable of running all tests = 0.5 x (2457 + 2656) = 2556.5
3dMark Score = 2.5 x 1.0/((1.7/2556.5 + 0.3/2409)/2) = 6333.1 = true

for Beanie:
GS for hardware capable of running all tests = 0.5 x (3823 + 3625)
3dMark Score = 2.5 x 1.0/((1.7/3724 + 0.3/1194)/2) = 7064.6 = true

So you can see the formulas are correct.

Now, substituting Solo's cpu score in my setup, but not considering the positive effect on my SM2 and SM3 totals, I would have had:

3dMark Score = 2.5 x 1.0/((1.7/3724 + 0.3/2409)/2) = 8605.4


This way we can now use the formulas to calculate very accurately scores for given combinations, should all else stay equal, and only one variable is changed. I would estimate a score well in excess of 9500 using Byron's FX60 in my setup.

Nice, hey. Let me know what you think.

Btw, I did not design the formulas, they are courtesy Futuremark - I just found them for you.

This post has been edited by mrbean: 28 January 2006 - 10:15 PM

[jasonhk]

Yea, that looks correct Thanks Beanie, quite a nice thing to know

Your GFX cards and Solos FX60

[mrbean]

Ah, pleasure.

I have done a quick calculation, and with my deadstock 4800+ scoring 1800 in cpu score, See my 4305 single 7800gtx/4800+ compare, if I use that in my calculation, the score would edge up to 8023.

Nearly a 1000 points gain going from FX55 @ 3100 to stock 4800+ at 2400mHz.

I will be totally happy with this arrangement. As I said earlier, I should be getting quite close to 9k with the 4800 overclocked to around 3gig, and using Byron's chip should take me to 9500.

Let me see how accurate my observations are when doing the tests, I am suspecting the cpu is going to have some positive influence on the SM2 and SM3 test results/scores.

Any feedback? Ralph/Byron? Jason?

[solo]

Maths so early in the morning

You made me find an online math tool but here it is

Take Beanies graphics cards, put them in CraigB's Dual Dual Core Opteron overclocked slightly to 2.590 GHz and .......

2.5 x 1.0/((1.7/3724 + 0.3/3567)/2)

9248.94

[mrbean]

Ouch....Byron, you by any chance have CraigB's addy details for me, and a time he won't be home?

I need to do some affirmative shopping on his setup, and Guys, I have 1st dibs on this one! Rofl....

Anyway, repek, those are very good cpu scores. Fokkit, seems the more you spend, the more behind you get. Eish. Let me see when I come back from OZ on the 12th of Feb if I can bring something nice with......

But a good tool to have, these formulas, very nice way to compare what we can expect with certain hardware configs.

Now we should be able to assist noobs and experts alike, and provide much better info when it is required.

Now, it would be a daunting task to build an accurate Database, b ut it would be nice to determine what weight certain cpu's at specific speedpoints add to the SM2/3 scores.

We all realize that a specific cpu should give a certain cpu score, but the effect this have on the other calcs.....mmmm

We need to build something like this.

I think what we can do, maybe Solo you can start, is keep gfx settings the same, but slowly increase your cpu speed, say 50mHz at a time, from 2600 up to 3300, and note your scores every time.

This way we can start building up a db for say, X1800's, X1900's, I can do the same for single 7800GTX/SLI 7800GTX's etc.

I think we would be doing the overclocking community a huge favor?

You game? Anyone else wants in, even with 6600's and up?

[solo]

I've got 2.6 / 3.0 / 3.3, but I guess I could do rest sure

[mrbean]

Thanx! Appreciated.....

[solo]

Beanie ... just for you

FX60 + ATI Radeon X1900XTX (650/1550)

3DMark06 - 1280x1024

FX60 - 2.60 GHz (200x13) - SM2.0 2289 - SM3.0 - 2451 - CPU - 1911 = 5719
FX60 - 2.70 GHz (208x13) - SM2.0 2297 - SM3.0 - 2459 - CPU - 1992 = 5777
FX60 - 2.82 GHz (216x13) - SM2.0 2302 - SM3.0 - 2467 - CPU - 2062 = 5825
FX60 - 2.90 GHz (208x14) - SM2.0 2298 - SM3.0 - 2473 - CPU - 2142 = 5863
FX60 - 3.00 GHz (200x15) - SM2.0 2309 - SM3.0 - 2470 - CPU - 2207 = 5901
FX60 - 3.10 GHz (207x15) - SM2.0 2311 - SM3.0 - 2488 - CPU - 2275 = 5950
FX60 - 3.20 GHz (214x15) - SM2.0 2316 - SM3.0 - 2497 - CPU - 2352 = 5995
FX60 - 3.30 GHz (220x15) - SM2.0 2335 - SM3.0 - 2491 - CPU - 2413 = 6033

Anything else ?

Need to take FX60 out of Mach II soon so I can try in on another board

[mrbean]

Solo, thanx for these, woulod be good for now

I need to dump my 4800 in the system, and I will do single and sli scores.

This way we can start building a db

[rra]

Quote

On 4800 with gfx settings the same in all benches 722/800

cpu at 2400 = 4530

2600 = 4583 +53

2750 = 4621 +88

2903 = 4681 +148

2903 = 4705 ( LOOSENED RAM TIMMINGS )

So looking at that it would be fair to say on dual core you gain only +/-50 pts for every 200mhz not much when you consder the cost for those 200mhz .So in summary I dont think cpu is as big a factor in this bench as the score for cpu.Overclocking GFX gives much bigger gains.

This from other thread might help

This post has been edited by rra: 30 January 2006 - 07:51 PM

[s e t h]

and here's my contribution

download here -> calculator