NVIDIA GeForce GTS 250M DDR3 vs. NVIDIA GeForce GTS 250M DDR5

Concerning the review of the Toshiba Qosimo X505 850 with the NVIDIA GeForce GTS 250M DDR5...

The review revealed 3DMark06 results of... 8,313
http://www.laptopmag.com/review/laptop/toshiba-qosmio-x505.aspx?page=2

Now compare this to the recent review by Notebookreview of the Acer Aspire 8940G with the NVIDIA GeForce GTS 250M DDR3

The review revealed 3DMark06 results of... 8,140
http://forum.notebookreview.com/showthread.php?t=434902

Am I missing something here? The DDR5 doesn't seem to be "far and away" much better than DDR3 like some people have claimed.

 

your basing the cards performance on one benchmark?

 

I'm sure I saw a benchmark for the GTS 250m GDDR5 that scores 9800+ somewhere.... here http://www.pcmag.com/article2/0,2817,2353367,00.asp

That is what benchmarks are for, though it is only an estimation of how the real performance is.

 

Hm, i already seen a benchmark of this card ON THIS site and the person had 11,000. So i would not believe that one bit.

I think that was in the turbo i7 mode. Many people also forget how cool these things run.

 

The DDR5 will be much faster in actual games. (min, avg fps)

 

wow,
that score's damn high for a 128bit....
need benchmark details....

 

nope it isn't GDDR5 effectively makes the GTS250M into a 256bit GDDR3 card... so the scores are normal...

 

We'd need a link to the test run here on NBR taht got 11,000 though :s

Otherwise, the OP's benchmarks seem ok at best so we'd need another opinion because it does look like they benchmark at similar scores, but benchmarks aren't everything(ex: the HD4670 benchmarks near the 9800M GS).

Also...

What resolution was the test run on though?

GDDR5 is better than GDDR3 because it doubles the memory transfer rate, effectively doubling the amount of data transfer available to the GPU. However, this transfer rate is most apparent at higher resolutions and on more texture hungry games. If the tests aren't run on anything over 1680*1050(or even over 1440*900), I understand that the difference wouldn't be as notable.

 

Just one more example of 3Dmark06 being outdated. It simply does not provide a very accurate measure of game performance. Vantage GPU score is not perfect, but much better.

Default res for 3dmark06 is 1280x1050, and since both of these laptops had 1080p screens, I would guess that is the resolution the test was run, but Laptopmag doesn't seem to document the resolution anywhere. I'm fine with assuming default res on these two, but Laptopmag's review of the Gateway FX 7808 has a 3dmark score that I strongly suspect to have been run at 1024x768, since it only has a 1440x900 screen.

 

Do you base all your opinions on ONE review/benchmark?

I am trying to say that you need lots of benchmarks (most should be REAL games, not synthetic ones)

 

And you can't ignore all the other factors such as CPU, memory, etc.

 

imo, the
gts250m/gts260m gddr5 scores shouldn't be too far from 9800mgt ,

 

In fact, they should be almost (if not) identical...

 

Maybe... in theory. The Toshiba also has 6GB DDR3 1066MHz memory. The Acer and Toshiba both have the same Intel Core i7 720QM cpu.

The resolutions IDK but given the similar results I would say they were tested on the same even if not the native Full HD (1920x1080 for Acer and 1920x1200 for the Toshiba).

The Toshiba is also like $550 more so at that price I would expect at least an NVIDIA GeForce GTX 260M.

And if DDR5 really pays off at the higher resoultions will the max at 1680x945 for the lower end Toshiba Qosmios really payoff over the DDR3 cards?

 

Also, I don't remember but I believe the DDR5 version in the Toshiba is actually underclocked, which would explain the lower scores?

 

Uhmhu. If it is a notebook I don't care about, I would not check more than one benchmark, unless I have my doubts (like in this case). For notebooks I'm interested in buying I do check game benchmarks.

Does GDDR5 tech really not increase 3Dmark06 scores?

 

Didnt you know? THIS...IS...TEH...INTERWEBZ ( insert front kick here ), which basically means that, unlike any really study or analysis, one benchmark ( now matter how flawed or how far from theoretical value ) is more than enough data to make a valid statistical claim.

 

At the end of the day, synthetic benchmarks aren't what you're using the GPU for; gaming is(or at least it's most likely to be). Therefore, the gaming performance is what matters.

 

My bad friends, i made a slight miscalculation. It was actually 10,400 and it was a BARELY overclocked GTS 250(Like 100MHZ) For those of you who are interested in checking it out go to the Official X500 Lounge in the Toshiba Forum.

You know what, here is a direct link:
http://forum.notebookreview.com/showthread.php?t=428805&page=8

 

Oh on a side note 3Dmark Vantage was about 4500 which (As shown in link) is pretty good.

 

4500s not so bad, its a good solution for nvidias mid range card as they call it, and the power is godly low too.

 

I know the power is low, cant wait till someone overclocks it to the limits to see how far they can take it.

 

Yeah definately, the voltage requirements though, i still cant get over the information i saw, unbelievable haha.

But its a really good performer, and i also bet its crazy cool too.

 

Since when is 20% barely overclocked?

 

My Radeon 5770 is 128bits and still eating a 8800GTX for snack

But as said, GDDR5 makes the difference in high resolutions. A 256bits card does not have the advantage in 1280x1024 let's say, same for GDDR5