Mobile HD4870 vs GTX280M speculation

sorry i am at high in the benchmark i just looked. i have to re-run those. there pretty old when i didn't know how to use this notebook right. i will redo them later tonight. i have 22 on high with a single 3870 now.

 

In many cases, having them increase the minimum FPS, giving a higher average.

I think there is very little difference between 3870 and 4850 in 3Dmark 06. Futuremark vantage will show 48x0's increased performance more accurately.

 

Also, try ATi Crossfire Xtension on Crysis to see if there is any difference:

http://epichardware.com/index.php?topic=12.0

 

no i was talking about crysis. no in vantage i am no where near those cards.


ok i will try that but how would it work when i open up the benchmark. wouldn't it start normal unless i used the program to open the benchmark and the crysis???

trying it tonight not home. on treo 800w.

 

Yup. The 4770 is a 4830 that saw his 256bits GDDR3 changed for 128bits GDDR5. The 4860 is mobile 4770 while 4850/70 are based on the 4850 right?

 

HD 4770 is the 40-nm RV740 GPU. It's the same 40-nm GPU that's used for the Mobility 4830(uses GDDR3) and 4860(uses GDDR5).

The desktop HD 4830 has been discontinued/replaced with the HD 4770 but it was a toned-down version of the RV770 GPU used for the HD 4850 and 4870.

 

Dont take my earlier post wrong, not every nvidia endorsed game sucks for ATI, infact Crysis is the only one I really know of.

They are doing great in every other game so far, ATI was beating Nvidia in DMC4 last I checked, and was on even ground in the majority of other games.

I have tons of 4870 Crossfire results in my W90 Review but its not a fair comparison until we see SLI 280's - http://forum.notebookreview.com/showthread.php?t=370294

Well I take that back, once somebody has a GTX280 system with a similar cpu I can turn off crossfire and run just one card and would be happy to bench for you guys. My unit is out for RMA right now and wont be back for like 2 more weeks though.

 

I keep seeing references about how fast the 4870's with DDR5 are going to be.

As far as I know they don't exist yet. Am I wrong? Is there a notebook out that uses them or laptop seller taking pre-orders for them?

I know we don't have 280's in SLI yet either, but at least the 280 exists.

I'm not loyal to any brand GPU, but do prefer the way Nvidia provides driver support for it's mobile customers.

 

Really how?

I know laptopvideo2go but that is not nvidia, those are enthusiast like you and me that mod the official drivers.

ATI doesn't have a dedicated website, but it has the mobility modder tool. Maybe ATI will be added soon since there are more ATI cards on market, back when laptopvideo2go came out it was all Nvidia cards.

 

I never understood this "Ati drivers sucks!" thing. I used ATi most of my gaming life specially since the 9700 triumphant era, and have never been disappointed by the drivers.

If something, I actually find VERY annoying the ammount of drivers I see from nvidia. I rarely see improvements until a third or fourth release, and sometimes it is not even worth it because it has a very small increas in fps.

But that's just me

 

@ ViciousXUSMC, Oh man, with mt M9750 w/7950's, I used to go right to the Nvidia site and get new drivers on a regular basis.

I like that better than the round about way we have to mob mod ATI's drivers.

Granted, they aren't tweaked, but it's as simple as download, sweep old drivers, and install.

You're right though, we'll likely be seeing better third party driver support now that ATI mobile GPU's are more common.

 

Crap, sorry for the dbl post.

For me, it's not that the ATI drivers suck. It's the fact that ATI doesn't provide mobile GPU drivers ready to go for the end user like they do for desktop GPU's.

 

Nvidia actually has the mobile drivers on their website now, check it out.

 

Don't blame ATI so much but rather blame the manufacturers that want to make their own low performing out-dated drivers.
PS. ATI has been putting out mobile drivers for Win 7 which have been working perfectly so far for myself. As well, once Win 7 is released, the majority of laptops will have Win 7, so drivers will be available for all.

 

Here's my opinion on the driver issue: As long as ATI can release a driver for the (hopefully) upcoming GDDR5 HD4870 that is up to par with the desktop driver, the hardware is what's going to really matter when comparing the 4870 and 280M, in which case the 4870 will win hands down. Theoretically this shouldn't be that difficult since ATI only needs to tweak the desktop driver to suit notebooks. It's not like they need to design an entirely different driver, am I right?

 

I'm with you Rob. Until the card is more than a blip on a PR sheet, it doesn't exist to me. I mean, all of this time goes by, and it still hasn't come out in any notebook? Plus, it's suspicious to me, that the Clevo M570ETU's manual lists a '4870 1GB DDR3'. Clevo, of all companies, would have been after the GDDR5.

I think the card was tested, but it's just too hot for any notebook that isn't built around it.

 

and clock speeds are importtant. if my clocks speed were the same as the desktop ones, i would rule the mobile gpus.

 

Lol, point being no Notebook card is ever as powerful as it's Desktop counterpart... I agree with Kevin that they probably tested it and found it ran too hot for any laptop that wasn't specifically built for it...to run it...

Although I would still like to try it in a Sager 9280

 

what about in a whitebook????

 

There's not a chance in hell 2 GDDR5 4870 would run in a whitebook without changing the thermal design. This is probably what they've been getting at since it's been so long since they made their promise.

 

Boom... you just got shot in the head..... ahh that was a bad joke lol

 

yeah we have been waiting and still are. hopefully flex will distribute them. other wise it's going up or sale.

 

But I've got my hard hat!

 

hummmmm. i just have one question.
what game is crossfire/ati single card good at?
and we all go from there. about tired of everyone bring up what games it's NOT good at.
matter of fact. the best game it's GREAT at would be nice.
then let the 280m go up against it.

 

Source Engine games are probably the most fair when evaluating ATi and Nvidia graphics cards.

 

originally taken from here...
http://en.wikipedia.org/wiki/Category:Source_engine_games

so one of these games then..right?
Pages in category "Source engine games"

The following 27 pages are in this category, out of 27 total. This list may not reflect recent changes ( learn more).

Template:Source modifications Template:Source engine games C Counter-Strike: Source The Crossing (video game) D Dark Messiah of Might and Magic Day of Defeat: Source The DinoHunters Dogfights: The Game H Half-Life 2

H cont. Half-Life 2: Episode One Half-Life 2: Episode Two Half-Life 2: Lost Coast K Kuma\War L Left 4 Dead P Portal (video game) Postal III S Salvation (video game) The Ship (video game) Shootout! The Game

S cont. SiN Episodes Source (game engine) Sting: The Secret Operations T Team Fortress 2 The Kill Point: Game They Hunger V Vampire: The Masquerade – Bloodlines Z Zeno Clash

 

in left for dead with my dual core t9400. i averaged 90fps everything including aa maxed out @1920*1200. i never tried might and magic before but i have the game. so i don't know the results of that.

i thimk dmc4 is the best because it doesn't require the cpu on it's benchmark and i think it uses crossfire efficiently

 

hooo hummm people just want to squish my dream on 4870's

ATI's driver support is 100% better then previously

just face it ATI is just as good as or better then Nvidia(current Gen)

 

DUAL CORE RUNS. THESE ARE OLD RUNS. IN FC2 I CAN GET 10-15FPS EASILY NOW IN DUAL CORE MODE.
here is a single run with a t9400 everything was at 2.53 unless other wise stated.



here is crossfire.


here is a single farcry2 run


here is crossfired.


a more up to date run of farcry2 with my quad but in dual core @2.53



RIGHT NOW MY COMPUTER IS OUT OF COMISTION BECAUSE OF COPPER MODDING. AND HEAT ISSUES. SO I CAN RUN ANYTHING RIGHT NOW.

here is 1680*1050 crossfire


here is overclocked


sorry i don't have single runs for that.

quad core qx9300 @3.2ghz. and optimized but stock 3870 crossfired



WHY DID I POST THESE RUNS?
I LOVE BENCHMARKING AND TO SHOW HOW WELL CROSSFIRE SCALES. ALSO JOHN YOU CAN RUN TO SEE HOW WELL SLI SCALES AND ANOTHERGEEK CAN YOU RUN THESE ALSO I KNOW YOU HAVE A SINGLE CARD. BUT AFTER WE GETS THESE, WE CAN GET A PERCENTAGE FOR THE 280M SLI. SINCE EVERYONE IS DYING TO KNOW.

 

It's more likely a supply issue than a heat issue as GDDR5 uses less power and produces less heat than GDDR3/4.

Qimonda, who supplied the GDDR5 for ATI's 4000 series reference cards, filed for bankruptcy in January and Samsung didn't launch their GDDR5 modules until February....and none of that worked with ASUS's plans of rushing the W90 to market in March with it's buggy BIOS.

So really what you refer to as "all of this time goes by" has only been 3 months.

 

we thought it was a heat issue?

we can i find it says it's has all thiss pros.

wouldn't ati not have alot of desktop cards also. unless they have alot of stock.

 

Sorry I'm not clear on what you just wrote don.

If you're asking about the heat issue keeping the 4870s out of a Whitebook it's because of the GPU chip itself not the RAM.

 

oh ok. i understand now.

 

Here you go, both overclocked @1680x1050. I still want a QX...

 

Nice! That's some serious power in a single card I love's it!

 

Yeah those are some really nice numbers for a 280M, it's approaching Crossfired 3870's. Anyone have a 9800M GTX to compare DMC4 fps?

 

Also remember he only has a Dual core at 2.3ghz I believe, where as The Tested M17 system has a Q9300.. Which means the 280m is probably just as powerful if overclocked (I know Dondahdah does) Do you Anothergeek?

 

cunfused? dmc4 benchmark's is just for the gpu. the cpu doesn't matter at all. the runs i put up where when i had my t9400 @2.53 dual core. and when i placed my quad at 3.2ghz it jump about 1 or 2 fps. that's why i like dmc4.

john that run is a single 9800m gtx stocked or overclock? also are you running 4 cores or 2.


also yes the 280m gtx overclocked is very close to the 3870 crossfired stock.
it is almost the same. nvidia did good in making that card, i must admit.

 

over clocked 2 cores.

 

i wonder how much i'm putting out on a single card overclocked. i never really thought of trying it. it maybe child's play compared to yours.

but i noticed i'm not at 1680*1050. i'll try to run it. i have to play with my driver first

 

i just ran it to put something up.

 

Lol. i see. i'm wanna run something also it's been about two days since i ran something.

 

off topic
lately i have been going for the single gtx280 (desktop) on air record. up to 23k for 3dmark06 and 12,940k for gpu vantagemark. gotta love these i7's though...

 

oh yeah i remembered you were getting a desktop i7. what's the cpu?

you stay setting records. Lol.




also i just remembered why we were putting up benchmarks, it's to see how the 280m gtx would do in sli. so john by providing your single and sli runs we can get a percentage for the 280m. and we can vs it versus the w90. to see the results.

 

Those scores from the 9800M GTX and GTX 280M are extremely close... but since both were overclocked, I think we should create some rules for benchmarking. Resolution will be 1680x1050. Either run them at stock clocks or overclock the cards as far as possible while maintaining stability. That way we can compare stock performance as well as max performance. Also, if you have a significantly slower CPU than the other people (200mhz or more), overclock it to match if possible and you feel safe doing so.

 

understandable.

 

Eh, even if he's running dual core he's still got ~333mhz x2 on me. Accounting for that, he'd still have ~39 fps, which is within the whole 10% faster idea.

 

Then I guess the easier thing to do is underclock, which is extremely easy with RMClock.

 

who me or john. i can downclock as much as 1.6ghz via bios. and i can lower my fsb if nessary.