Radeon HD 2600 vs GeForce 8600M

In a revision of a previous topic, let's see the rundown. The ATI Mobility Radeon HD 2600 vs the nVidia GeForce 8600M. Let power not be all that is counted. Smoothness of operation, ease of drivers, power usage, and all such other factors should be included.

 

smoothness of operation?

Honestly if we can't even get a clear 3dmark indication of the HD2600 I can't see how were going to do a performance test such as this. Or find anyone who has the laptops and would be willing to.

 

If you tell me what that is, and i like it, once i get my computer I will do it.

The HD2600 GDDR3 is THE SAME (3dmarks wise) as the 8600 GT OKAY?

 

Performance-wise -
According to one review (of a Samsung laptop, I believe), HD2400 gets noticeably worse FPS than 8400GS in games it tested, even though HD2400 has slightly better 3Dmark06 score - so yeah, I agree that we need some real-world benches, not 3Dmark scores.

Also, despite of what above poster says, HD2600 and 8600GT have significantly different hardware architectures, so they tend to perform in different manners, even though DDR2 8600GT and HD2600 shares similar 3Dmark06 scores.

About the OP's other questions, I suspect that HD2600 may consume less power than 8600GT, since it is on 65nm. Ease of use should be same for both GPU under Windows. However, HD2600 still has no 3D support in Linux.

 

Comparing the Radeon HD 2600 GDDR3 to a 8600GT DDR2, performance should be equal although the Radeon has a slight performance advantage. If you compare the Radeon to the GDDR3 GT variant then the GT should be better.

 

There's no way to know that, as we have not yet seen any stock GDDR3 HD2600 lappies in the wild. The 8510p performs for all intents and purposes the same as a G1s or MBP at the same clock speeds.

 

I can't remember what the 8510p is equipped with, but my earlier comment will have to be revised in light of your conclusions odin243.

 

what is the 8510p equipped with?? i believe the ddr3 version correct??

 

It's an underclocked GDDR3 at 450mhz mem. If clocked back up to 600MHz mem, it would almost certainly perform on par with the standard GDDR3 8600M-GT (of course nVidia will always be better in some situations, and ATI will always be better in some situations.)

 

this is one of those threads that needs to be stickied. seems more and more people are asking the same 2600 vs 8600 questions. they're about on par, from what i heard, except for haven't seen any benchmarks...

 

They are on par on 3Dmark06, that does not mean they are on par in real-world settings (there's no benchmarks seen aside from 3Mmark). We cannot draw the conclusion only from 3Dmark06 scores, especially when we know HD2600 architecture favors 3Dmark. Despite many people complain about this, it is much better to estimate Mobility HD2600 performance from their desktop counter parts rather than from its 3Dmark score. Desktop/laptiop HD2600 has the Exact SAME technical specification- transistor counts and such - on ATI site. And they tend to to perform fair bit worse than 8600GT clock-per-clock, outside of 3Dmark.

HD2600 and 8600GT aren't very much like each other. HD2600 is more like 8600GS/8400GT (in term of numbers of texture backend, etc) with shader power that match the last-gen high-end hardwares (much more than 8600GT). That means, sadly, HD2600 is often short of raw power to utilize it hsader power ATM. All in all, I still strongly recommend 8600GT over HD2600, DDR2 or DDR3 - however, I do recommend HD2600 over 8600GS/8400GT.

 

agreed. 3dmark was never an indicator of anything except an ego satisfaction benchmark. we'll just have to wait for an actual user to post some benchies... PLEASE lol

 

No, those conclusions are entirely misleading, as ATI does not publish full technical specs on their site, and no one knows exactly what cutbacks ATI makes to their desktop cards to turn them into laptop cards. In every previous generation the ATI vs nVidia laptop mid-range comparison has been entirely different than the ATI vs nVidia desktop mid-range comparison, usually favoring ATI on the laptop end and nVidia on the desktop end.

The only difference between the 8600M-GS and the 8600M-GT is the shading power difference. As you've agreed that in that respect (shading power) the HD2600 is similar to the 8600M-GT, I fail to see why you're comparing it to the 8600M-GS.

And for the record, there are more HD2600 benchmarks than 3dmark06 available now, there's FEAR and CS:S most notably, which agree that, clock for clock, the HD2600 is on par with the 8600M-GT.

 

Yea the F.E.A.R and CS:S comparisons we've seen have been pretty much equal. (im too lazy to find links just use the search function). Although arguably F.E.A.R and CS:S are good ati benchmarks.

 

I've got an HD2600 (Toshiba A200 TH7) so if you can point me to some tests I'll run some for you guys. (Note they must be free because I don't want to buy a bunch of games to test)

 

run any games you have...the newer the better.....or download the Bisoshock demo...pretty much any demo for any new game. See how it runs, download FRAPS and see what kind of FPS you get during the newer games. Hell, even running doing this on ONE game would be helpful, thanks.

 

also, download the World in conflict demo, WITH the latest and most stable drivers

 

I did do the Bioshock demo and unfortunately something is messed up, as it keeps freezing after I go down the elevator to rapture. But before that I could get ~20fps at 1280x800 with all settings maxed except the 1st 2 (windowed mode and something else that's useless). Also note, using the 7.9 CCC modded with mobility modder and all settings default.

 

That sounds good. You have the DDR2 HD2600 probably (only the HP 8510p seems to have DDR3), so DDR3 HD2600 should be even better.

 

But that can change with an OC, just waiting for new version of ATI Tool. And FTR I score about 5800 on 3dmark05 and 3000 on 3dmark06.

 

Yeah, that sounds pretty good, especially considering ATI driver support for the HD2600 is very new (I'm assuming that's why it crashes), nothing near as good as the 163.xx drivers for the 8600M-GT.

 

Yeah, but it only crashes at that particular spot and not necessarily at the same spot, sometimes I get a little bit farther, sometimes not. I'll get onto a couple more games soon, give you guys some pics.

 

Did the WIC demo benchmark at 3 settings:

1280x800 - very low settings
Avg - 41
Min - 12
Max - 104

1280x800 - med settings
Avg - 12
Min - 5
Max - 29

1280x800 - very high settings
Avg - 3
Min - 1
Max - 8

And FTR, I could hardly notice a difference between the very high/low settings and min values happened when the bombs went off creating smoke clouds (the worst thing for fps), I will also post some results on the crysis beta which is now out on fileplanet (ie, highest playable res and detail/ lowest res and detail)

 

8700GT gets ~27FPS on medium , not sure what the 8600GT gets. 12 seems pretty bad, well then again not TOO bad considering but I was hoping mid-range could play latest games on medium.

 

I would attribute it to mostly immature drivers and the fact that you're matching up a "different" class of performance video card. But due to the 3dmark05/06 scores, the (G)DDR2 HD2600 matches almost perfectly with the (G)DDR2 8600M GT, now the HD2600XT (GDDR3) matches up with the GDDR3 8700M GT which you're talking about.

 

3dmarks are not a good indicator of performance differences between nvidia vs ati. Ati notoriously scores equal or higher in 3dmarks only to flop in in game performance. Main use for 3dmarks is just to get an idea when comparing drivers for the same gpu.

 

Ok, didn't know that, I guess you can still compare Nv to Nv and Ati to Ati right?

 

ati is generally good for games that require massive amount of 3d rendering at once but are slow paced. nvidia performs better at games that need good fps...generally speaking.

 

Very similar results for my 8700m...

the low settings is because cpu hit 100% on mine...
Seems 8700m is quite easily limited by its CPU,
or we need better drivers to minimize redundant cpu-gpu traffic!!!

 

Umm no - shading power isn't the all difference between 8600M-GS and 8600M-GT. 8600m-GT has nearly twice the fill-rate of 8600GS, which is quite a bit different from mere shading power (and much more substantial). 8600GS/8400GT/Desktop 8500Gt has half the ROP of 8600GT (8 to 4), which means it has half the raw power of 8600GT. HD2600 also has only 4 ROP - which is often identified as the reason for its weak performance.

Now, back in the last two generations (7600 vs X1600, 6600 vs X700), I do know ATI mobile products perform comparable to competing Nvidia products, despite they performed much worse than Nvidia with equivalent desktop cards. However, in last two generations, Nvidia midrange mobile GPUs had considerably slower clockspeed to their desktop counterparts, whereas ATI mobile GPUs are clocked closer to their desktop counterparts , enabling them to fare better in the field of mobile GPUs. This isn't the case with this generation - mobile 8600GT is only slightly slower than desktop 8600GT, assuming it has DDR3. Whereas, there is no way mobility HD2600 is going to be magically faster than its desktop counterparts.

 

Okay, if you differentiate between shading power and fill rate, then you're right And to be clear the 8600M-GS has 8 ROP's just like the 8600M-GT. However it does have half the TMU's (8 vs 16). And also, fillrate is increasingly less important compared to shading power in newer games.

Clockspeed was not, and is not now, the full story, nor is it sufficient to be able to tell how mobile cards differ from desktop cards based on clockspeed alone.

And if everything you say is true, why does the MR HD2600 perform almost identically to the 8600M-GT at equal clock speeds?

 

Fillrate is still important for mid-range card though, as they don't have enough of them (yet). Otherwise, HD2600XT GDDR4 would have trashed 8600GTS.

Also, laptop 7600 only had 8 pipelines, down from 12 of desktop 7600GT, which cut down their performance fair bit. And I thought HD2600 at 500/400 perform slightly worse than 8600GT at 450/400 (Dell inspiron versions), even in 3Dmark06?

Hd2600 does seem to overclock exceptionally well, though.

 

I really can't believe we have yet another thread on this topic

 

does the ATI 2600HD support XP and Vista or just Vista?

 

True enough, I mainly meant that on newer architectures it's closely tied to shading power, so I usually don't look at it seperately.

Yup, that's one of the things I was referring to when I said you can't only look at clock speeds.

The 500/400 HD2600 seems to perform slightly worse in 3dmark06 than the 475/400 8600M-GT, however the only real world benchmarks I know of are from the 450 mem HD2600, which seems to perform (in games at least) significantly better than the 475/400 8600M-GT, though slightly below the 475/700 8600M-GT. I believe stock speeds would be 500/600 for the HD2600, which I have no doubt would be extremely similar to a 475/700 8600M-GT.

I think that's because the 8510p comes factory underclocked, I haven't seen a stock clocked HD2600 in the wild yet.

 

Crimson: It supports both

I'm confused though...If people here did their homework they'd know the underclocked HD2600 (in the 8510p, for example) gets around 3300 in 3dmark06, whereas the "Dell" DDR2 8600GT gets 3000. Furthermore, the GDDR3 8600 GT found in the Asus G1s, T61p, 8510w, etc, gets around 3600-3700.

So really, the underclocked GDDR3 HD 2600 is right in between the two. Return it to stock speeds and you'd probably get something near the GDDR3 8600 GT.

But wait, 3dmark scores don't matter right?

 

Ok guys tomorrow I am getting a Toshiba A200 with a ATI HD 2600 with 512mb (not 256mb like the previous user). If the user is able to get 3000 in the 3Dmark06 score, what should this card be able to get. Also is the 512mb version of this card better then a 8600m gt ddr2 512mb?

plowking

 

It's real similar to a 256mb DDR2 8600M-GT, and probably slightly less powerful than a 512mb DDR2 8600M-GT.