Mobile HD4870 vs GTX280M speculation

While I have scored higher than this, this is the best I have uploaded.

3dMarkVantage Performance

Overall: P10451
GPU: 10421
CPU: 10543

CPU: Q9000 @ 2.745ghz
GPU: ATI 4870X2 (4870M Crossfire)
Asus W90VP-A1

 

Here are mine

From W90 Benchmarks

Ran at 1920x1080 everything maxed out.

Your score @ 2.66ghz 1920x1080
105
83
125
76

 

stocked right or you are overclocked?

 

Overclocked on Vantage as that was after my review, the DMC4 is stock as I reviewed the laptop at stock performance.

However I re-did vantage with gpu overclock and it made very little difference except for extreme with maxed out custom settings as I am cpu bottlenecked to extract more power out of the gpu units. They are faster than the cpu except for 1920x1080 w/ very high settings, Performance Test uses 1680x1050 with medium settings,its simply too easy for the dual 4870's with only 2.7ghz of cpu power to render with.

So you can consider those stock results.

 

Here's a 1080p run from earlier.

105 83 125 76

 

Isn't it possible to run 1680x1050 with max settings on DMC4?

 

Cool edited the above to compare it with.

 

I can do an AA run I guess... but I'm lazy right now lol

 

found
http://i39.tinypic.com/2mf0l1i.jpg

 

Oh, and for the record, I did have a Crysis 30 FPS all high run at 1080p but I don't think I saved it <.<

 

That is an impressive score. With my desktop overclocked with CPU at 3.6ghz I also got between P10000 and P10500. This bodes well for the mobile GDDR5 version.

 

you sure about that....
i still have the vantage dual gpu records for Performance/high & extreme

at least till the sli 280m's surface....

 

Just GDDR3 on the Asus Cards, I guess you meant that this means the GDDR5 will be even better, but that we have to wait and see. As stated my cpu is holding the gpu back right now, even with faster gpu my score would have been the same.

 

it's not your cpu...it's really your fsb...kind of why they up'd it for the i7's from 1600mhz to 3200 mhz and up...(personal opinion)

 

i7 is a different structure you cant compare the same fsb speed to an older series, but even in the core seriews the fsb is way faster than anything else in your system, its just the system bandwidth, much the same way SATAII has a XXXXXXX speed but the hdd can only use XXX of it.

 

tis is true...
but fsb also controls memory speed as well.. and if your memory can't keep up since it's where your game is loaded...driver display crash...(speculation of course)

 

Well this also goes against all my personal experience. I have done ram only overclocks on my desktop before on several occasions and even the highest overclock never gave more than a single fps gain in games, and the only benchmarks it was apparent in was a very small wprime/cpuz boost and of course memory bandwidth tests.

But all my experience points directly at cpu bound benchmarking, when a gpu overclock shows no gains but a cpu overclock does, the cpu is holding me back.

The doing a custom performance run with every option at its highest setting, the gpu overclock suddenly made a big difference (over 300 points) because it was so hard to render the gpu were under enough load to eleminate most of the cpu bottleneck.

Be it FSB or Raw CPU speed, one thing is for sure the bottleneck is being relived from overclocking the cpu, and the cards have more potential than I can currently able to extract.

Did you ever do a maxed out performance test? My GPU score was like 5340 or so with gpu overclock and without it it was like 5000 even. I dont have those benchmarks uploaded and they were on my W90 which is now RMAed lol, but I have them in an acronis image so I can get them if we really need them.

 

never did the maxed out performance test, only the maxed out extreme and high test we ran a while back....

wish i had cpu over clocking on this then we would see the true speed of the 9800m gtx. since over clocking cpu seems to be a heavy influence on every ones test but mine. this is all done by the gpu over clock only...no other enhancements. and going by what your saying...then i should easily do about 19k with little to no trouble what so ever... or 12k to 13k gpu vantagemark....

but also given what you explained should make it pretty clear to others...if the np9262 had cpu over clocking...im gonna have to say it would be pretty high up the list of benchmarks and gaming benchmarks and gaming, but yet..others seem to not think so.... like it would be completely impossible for this out dated/rebandage gpu to keep up...yet it does....

what are the settings for the maxed out performance?

 

I used the Performance resolution but went into custom settings and raised every single setting to the highest possible, its actually even higher than extreme for the AA or AF I forget witch.

 

Now that I think about it (from the last debate we had) I think you are confusing the FSB with the PCI-e Bus/Bandwidth, which is not in-sync with the FSB in current chipsets. This means that upping the FSB won't up the PCI-E bandwidth. Looking back at my results as well as Vicious' runs, SetFSB showed no change in the PCI-e frequency when the FSB was increased. The version I used, however, had a separate control that allowed me to control the PCI-E frequency. Upping it to 110mhz only got me back to the GPU score I had when I just upped the multipliers on my system.

So you won't be seeing an 8-20% performance increase (12k or 13k Vantage GPU score) on your system even with a FSB boost. If anything, a PCI-E boost might allow you to clock a bit higher, but not by much over the previous best.

I had a similar experience too when fooling around with my HTPC, but noticed games like GTA 4 and C n C played smoother after getting my memory to 1800mhz.

As for FSB controlling memory speed, I think it does, though asynchronously (which is why you see some systems with newer CPUs working with DDR2 RAM).

 

Well I did it, ordered an E0 ES QX9300 (QGNG) off Ebay for $400, along with some OCZ freeze and thermal paste remover fom newegg Hopefully it OC's to 2.93 fine lol

 

True, but if you are implying that the increase in memory speed translates to that 8-20% gaming and GPU performance increase you were talking about, then I would strongly disagree.

No, I agree with this. FSB overclocking is a cheaper and better alternative than buying an extreme processor, which as you mentioned, has the other requirement of needing a BIOS that allows unlocked multipliers.

Not saying that at all. The results show that for ATi cards, the GPU is heavily influenced by clock speed, which can be achieved both ways. A CPU running at a higher FSB wouldn't significantly add more performance in this respect over a GPU running with more multipliers at the same CPU clock speed. This is why when the Q9000 in the W90 was overclocked to roughly the same clock speed as a T9600, the GPU score of the W90 with a Q9000 @ 2.75 ghz was similar to a W90 with a T9600 at 2.8ghz. The T9600's FSB as that speed was 266mhz, while the Q9000's FSB was around 370mhz

And because nVidia cards aren't as bottlenecked by CPU speed as the ATi cards seem to be, you probably won't see anything significant in terms of GPU score from bumping your clock frequency. Overall performance, however may improve.

 

True, but if you are implying that the increase in memory speed translates to that 8-20% gaming and GPU performance increase you were talking about, then I would strongly disagree.

oh no!...lol.
we agree there.

only saying that it has a varying affect on system wide stability.

ok, lets say not more performance..how about a better tailored performance? maybe that's how i should have came at the whole situation? since you can move in 1 mhz increments while multipliers move in chucks... is this how i should have said it way back when?

well.... i went from 18k to 23k...that seems to be a nice boost...lol (desktop i7 single vga)
i went from 14k to 18k over clocked (desktop quad single vga)

so i was trying to apply the same theory to the laptop for nvidia cards...

 

Well, chipset architecture in laptops shouldn't be that much different from desktops. But in your case, you are referring to an i7 desktop, correct?

 

both actually...

well.... i went from 18k to 23k...that seems to be a nice boost...lol (desktop i7 single vga)
i went from 14k to 18k over clocked (desktop quad single vga)

 

welll in my run with my t9400 setFSB i went from max 12,400 to 14,9xx i could of broke 15 k but not much higher

 

i really don't like 3dmark06. at 3.2 i got 15858. and 2.53 i got 13???, and at 1.6 i got 8500.