Nuclearus Multi-Core Benchmark

[Monkeywoman]

download from here; http://nuc-rus.narod.ru/eng.htm

homepage here; http://nuc-rus.narod.ru/index.htm

Benchmark use algorithms like:

For ALU
- calculating Fibonachi digits;
- calculating Factorial;
- calculating with formula Ferma;
- brute force(excess password);
- addition matrix;
- multiplication matrix;
- quick sort;
- putting non-attacked queens;

For FPU
- Hyperbolic cosecant;
- Hyperbolic tangent;
- calculating digit Pi;
- calculating Mandelbrot’s fractal;
- rotation planets;
- calculating factorial;
- Prime;
- calculating digit e

Multi thread tests use four ALU and four FPU tests.

http://valid.x86-secret.com/show_oc.php?id=421838
first run;

[NapalmV5]

[MomijiTMO]

Q6600 @ 3.6.

[Speederlander]

This is a 24/7 system, not a bench system, with Vista, no tweaks, simple BSEL up to 3GHz, normal processes running in background. Pretty sure I could squeeze more out of it just by shutting down some neededless stuff.

[[XC] riptide]

This is the original Thread here http://www.xtremesystems.org/forums/...48#post3134348 . It seems that it does not scale as well as it should do especially for 8 core behemoths like yours Speeder. Sure some of the sub tests do... but Digit E, Pict. Morphing etc did not.

[Speederlander]

This is the original Thread here http://www.xtremesystems.org/forums/...48#post3134348 . It seems that it does not scale as well as it should do especially for 8 core behemoths like yours Speeder. Sure some of the sub tests do... but Digit E, Pict. Morphing etc did not.

Yeah, I noticed that.

[NapalmV5]

This is the original Thread here http://www.xtremesystems.org/forums/...48#post3134348 . It seems that it does not scale as well as it should do especially for 8 core behemoths like yours Speeder. Sure some of the sub tests do... but Digit E, Pict. Morphing etc did not.

Yeah, I noticed that.

idk how efficiently nuclearus benchmarks multi cores but if you look @ fritz benchmarks:

http://www.jens.tauchclub-krems.at/d...enchmarks.html

same results: 4 x 5000 faster than 8 x 3000

so, speedlander if you want to break 30k nuclearus 8 x 3200 should do

[Cheffy]

there was a version release (check the orig thread) that better catered for many core systems. i believe it ran the test longer/multiple times. but the results should be comparible to 2.0.0 directly.(eg any cpu should get the same in both, only many core beasts gain from the tweaked one)

[jcool]

Yeah, seems kinda slow on my Octa



Only ran it once, maybe I could hit 40k if it fluctuates...

[NapalmV5]

are you guys happy now? you bunch of 8core babies (referring to the cinebench10 thread aswell)

[jcool]

Loooooooser
Now I'm happy

Oh yeah: Seen my latest CB10 yet?

[skulstation]

after cb10 hear ar my Nuclearus bench scores.

[xtant25]

E8600 @ 5337MHz.

[Jacky]

Would anyone be so kind as to post a nehalem nuclearus run as soon as the NDA expires or he/she is free to do so? Maybe even with smt and turbo on/off.

I shall update the post with my own q9450 score later.

[WVUEcoCarTech]

Comin for you Jcool, now that there are Vcore settings via BIOS. Run was about 1.08V and began to become unstable once in CB10

[[XC] serlv]

here's mine. this friggin system is getting hot! need better cooling.

[WVUEcoCarTech]

here's mine. this friggin system is getting hot! need better cooling.

what kind of settings do you have via BIOS? have you messed with any settings for RAM or anything. What types of temps do you notice with them OC'd that much and what type of cooling are you using?

Is your L1 L2 write speed dependent on bus speed?

[Patagonian]

C2Q Q6600 @ 4050Mhz vcore 1.6
Muskin 2x1giga ddr2 1125mhz 5.4.4.12
Thermalright Ultra 120 eXtreme + Tacens ice 12x12
GA-P35-DS4
21788 Points

[fellix_bg]

[carepolice]

this for start, will do 4000Mhz later today.
ram is 1:1 -> 5-5-5-18 440Mhz becouse aData sucks really hard, goes up to 500Mhz and after that stability is bull

[Madman666]

Can someone please post the results for an Intel 920 i7 @stock and @3.2Ghz?
Please please

[bowman]

No 3.2GHz, unfortunately, the stock cooler combined with crappy seating means this one's running at stock until the TRUE adapter arrives.

Apparently this benchmark doesn't know what hyper-threading is and thinks it's an 'octo core'.. Might have an impact on performance, I know it does in LINPACK and TerraGen.

[Madman666]

@bowman: Thanks !!!

[luie]

Cool benchmark. I like how they name what they are testing.

[Mechromancer]

No 3.2GHz, unfortunately, the stock cooler combined with crappy seating means this one's running at stock until the TRUE adapter arrives.

Apparently this benchmark doesn't know what hyper-threading is and thinks it's an 'octo core'.. Might have an impact on performance, I know it does in LINPACK and TerraGen.

What are your scores with HT off?