*OFFICIAL* M15x Benchmark Thread

so i had time for one bench yesterday before leaving work, and here it is. dang it was so close to 10k gpu. next time i will include a shot of gpu-z sorry.

 

Hey sorry i logged off quick yesterday and didnt get the chance to answer. busy busy day. anyways...,

yea i flashed my gpu with a modified vbios using RBE tool, and increased voltage to 1.15v. Was super stable. Stock clocks on the dell 5870m are 700/1000 just like the M17x (since this card isnt offered for the M15x, and was intended for m17x's only). ALso had a strong fan blowing air on my m15x, with ambient temps at 22-23*C. Temps never went over higher 60s if im not mistaken with that mother of a fan blowing on it ( its more of a blower than a fan).

And Inap, Thats a friggin high score for a 5850m! Would like to see u break the 10k though keep pushing after all, your card is under warrantry now hehehehe

 

hahah thanks man. only had time for one bench and just picked a random clock. hahahha i will see if i can get it to 970 or 975 core.

 

would u be brave enough to test out my theory? that is, to flash a dell 5850m, with a dell 5870m vbios? The cards are so close im almost certain that that's the only diffrence bewteen them, and would then allow 5850m's to OC just as high as the 5870m's...

Any die-hard OC-erz wanna test this out?

***TRY AT OWN RISK THOUGH*** Don't wanna be held responsible for bricks ... hehehe

 

hahahh probably not just gonna see how far i can push it with 1.1v first then gonna flash to 1.15v later.

 

lol didnt think so ... i'm positive you'll break 10k easily if you didnt even flash it to 1.15v yet, and got those scores... hell you'll probably beat my score too

 

look who just joined the 10k club. yeah!!!!!
still get a feeling that my 920 is pulling power away from the 5850. i think the gpu score should be higher for these clocks.

 

awesome score now im thinking i might bench once or twice, once im off work tonight.

 

10K Club AND 19K Club -- together. Thin air up there. That poor 5850 is going to disintegrate if you're not careful. CONGRATULATIONS, Inap. My hat's off to you.

So, how high was it on that last run?

 

it failed at 990 core, but oh well still on cloud 9, hahhahah

 

"Failed"?? @990. LOL. Did TDP get raised?

 

Lol yeah failed on new calico in the strangest spot,right at the end of it. Still using 78 for tdp. Man but 10 core clocks behind the best 5870 is pretty good. I'm way happy with this card. Even when it dies I will mount it on a wall or something.lolz.

also the strangest thing re ran at same 985 core and got a way lower score. i have an i3 right here wonder what that would do.hahahha but maybe later, gonna go out now and enjoy the rest of the day.

 

dell like to receive their parts back within 10 days or they start leaving hate mail on your phone and start calling collectors.

 

Dude, that happened to me when I first hit 900. It was too low of a voltage though.

 

Yeah seems its all random,I can run the same clocks twice and get 2 way different scores.

 

Woooooo Congratz bro that's friggin awesome! See I knew u'd make it and I'm pretty damn sure that Dell 5850s are 5870s... Even if they're not at this point, does it matter? Inap ur clocks n score prove that there's no diff between them performance wise... Like zero... And 10 clocks behind my card can simply come down to the card, as not all cards oc the same. Anyways all this to say well done and ur due for sum rep bro

 

After having to call back to have them replace my 5850 for the second time they put me in touch with one of the actual hardware techs. He said the 5850 is the EXACT same physical card as the 5870 except with a few shaders turned off software side to run cooler. And since they where so far back ordered on the 5850 they where going to try and replace mine with a 5870. They ended up not being able to do it because they put out more heat and apparently they dont make a spec heat sink that will fit the M15X.

 

The main difference between 5850 and 5870 other than clocks(which can be manipulated) is 5870 has more transistors(can somebody correct?) and shader blocks(no this is not a software side effect :/ I wish it was lol), which overall results in an increase about 0.12TERAFLOPS of processing power. 5850 has 1TERAFLOPS so 5870 has 1.12TERAFLOPS.. I just like the word TERAFLOPS

 

yeah i'm still real stoked about it, hahahahhah been in a good mood all day.lolz i was really surprised i could push it to 985/core. never seen anyone do that.

 

haha awesome bud. can't wait for my 5870 to get here. not gonna be pushing it to breaking point like you are though

 

Im by far no expert on gfx cards, but I think this might not be true. For what would be the transistors used? Transistors alone done buy you anything, they are used to implement logic gates (like and/or/xor gates), which make up the digital circuit.

As to the difference in teraflops, that seems about right. If you look at the core clock of 5850 its 625 Mhz and the 5870 is 700Mhz. Thats about 10% difference, which results also in the 10% on the side of the flops.

I just checked the ati website, and the hardware numbers seems to be the same:

ATI Mobility Radeon? HD 5800 Series Graphics

 

^^
i suspect as much

it may just be the batch test.

stress out each chip with a set of benchmarks, those which pass the highest series of tests get binned as 5870's while those which fail get put through a series of 2nd tests and if passed... binned as 5850's

as a result and due to supply and demand sometimes a chip may get swapped from 5870 to 5850.

imo the 5870 is an overclocked 5850

in the right hand's we've seen the 5850 is just as powerful... inap.. where are you? too busy benching i'd imagine

oh don't forget that initially the 5850 came with ddr3 rather than ddr5, we are assuming of course the ddr5 version.

 

Yea im right with you Xeroxide, on the stress test theory.

That would explain why the 5870's can OC just a bit more than the 5850s... but are otherwise equal.

 

I was checking Notebookcheck and was suprised with 720QMs PCMARK Vantage scores.

720QM
min: 3452 avg: 5504 (51%) max: 6392 Points
820QM
min: 5540 avg: 6128 (57%) max: 7392 Points
620M
min: 5575 avg: 7345 (69%) max: 10611 Points
920XM
min: 7408 avg: 8971 (84%) max: 10654 Points

Does anybody know why quads are getting lower in Vantage? I was expecting atleast 820QM to butcher 620Ms scores :S but there is a huge difference :S

 

The 620m is a dual core, and the clock speeds are WAY higher than each core on the quads. This means that programs that don't utilize all four cores, like vantage, are going to score quads worse than the duals.

 

We need more benches. It would be great if we can all do a vantage cpu run and post them. I would rather see real M15x benches and would make a good reference.

 

PCMarkVantage measures performance on common general use tasks such as viewing and editing photos, video, music and other media, communications, office productivity, security, etc. HDD speed is a major determinant of scores, while neither powerful processors nor high end video cards are particularly influential. The test is usually used for relatively low-end, general purpose laptops and netbooks. How did we get into PCMarkVantage?

If you want to compare processor performance, run wPrime v1.55. It would be interesting to see how the i7-620M does. I have only seen a few tests; HWOrg shows 13.9sec +/- as the best score on the 32M test for the 620M, but there are very few results reported. If we could get some forum members with 620M's to run the benchmark, we could get a better idea of how the i7's compare.

 

Here's mine.

 

Cool. Before you ran it, did you click on Advanced Settings and insert 4 in the box? Need to be sure all threads are firing.

 

here is another 620m test with wprime

 

Here's a wPrime score for the 820QM for comparison.

​

Because the 820QM has a TDP of 45W, it can't even run 4 cores/8 threads at the rated multiplier of 15x. It can only manage about 14. A 920XM, with a TDP of 55W, running with turbo-boost disabled, i.e. at 15x across the board, can complete the 32M test in 12.65 seconds. With turbo-boost enabled, with stock multipliers, it runs the test in 11.146 seconds:

​

With further overclocking, the 920XM can get down to about 8 seconds.

 

Yes, that was a four thread run.

 

so whats the recommended setting to run this?

 

For you Inap, it's 8 (4 cores - 8 threads). Show what that thing can do.

 

Yeah, Extreme chips are monsters with the ability to run 2-4 cores at 3.2/3.33ghz now 920xm/940xm). Just beastly.

 

sweet here is my 920.

tstop running but no oc or tdp change.

 

They really are. And with T-stop letting you essentially remove the TDP limits, wild-eyed, screaming monsters. Which makes talk of how other Core i7 processors are "nearly as fast' or "competitive" or the like just silly.

 

You forgot something. LOL.

 

How do you do that? I have been spending so much time on other stuff I am now going to mess with all of this.

Bin sorting, etc exactly how processors are made and tested. He is exactly right. Has anyonehere benched a 5850 (at 5870 clocks) vs a 5870 (stock)?
I haven't read this whole thread yet... I would be rly interested in seeing that info.



Lastly does anyone happen to have a L4D2 recorded demo I can use for benchmarking purposes? Thx

 

You mean how do you use Throttlestop in the o/c efforts? It's the silver bullet,

 

wait did i do something wrong,lol really new to this bench.


and yeah tstop is the ultimate 920 oc tool.

 

The link in your post is broken. Can't see the result.

 

oh hahahha can you see it now?

 

24x across? Still, awesome.

Even with 24x across, I only get 10.1 seconds on the Sager if I leave the TDP at 55W. Have to go up to 65W to get 9.6 sec, and even then the effective multiplier is only 19.8 +/-. What magic spell are you using out there?

 

damn good drag time

 

yeah 24x across. stock for the M15x

what is reg stock for the 920?

 

Yes thats correct.. not sure what the TDP stuff is or even how to OC on this. Well I am sure I can figure somethings out but I assume that guys already know the best things to do.

I noticed that too.. even in other things you were showing me.. remember that throttling stress test.. How are you doing that?

 

Regular Intel stock is 24-23-17-17 on the 920.

DeeX, we'll be happy to share what we have learned, understanding that it is still very much a work in progress. But there are some real experts around that are usually happy to answer questions.

 

Wprime on the 940xm. I actually got it down to under 7 seconds before but it BSOD'd when I took the shot. Need a bigger PSU for this M17x.

 

nothing just using tstop to keep it from throttling but thats it.
but i did 17k vantage stock for this cpu.