Page 2 of 5
Posted: Wed Jul 23, 2008 4:17 pm
by Kel Ghu
GDDR3 is different from GDDR2 in the sense that requires less voltage, therefore runs cooler, due to new production processes. So it is usually implemented with higher clockspeed. But at same clockspeed and bus width, it performs almost exactly the same (difference is due to page size, but its not noticeable).
So, if you say GDDR3 is faster due to higher clockspeed, yes I agree. It allows you to play games with larger and nicer textures and somewhat speed up antialiasing. But it won't change the fill rate capability of the GPU nor speed up shading processes. Games are much more often limited by the GPU and not the memory. Pimping up the memory without doing the same to GPU will be only marginally faster. People on those forums often talks about little details they probably will not see in real usage.
I can tell you that my overclocked FX570 is much fater than stock, but if I only overclock memory, there's just little difference in games.
Posted: Wed Jul 23, 2008 4:20 pm
by arlab
awolfe63 wrote:I think the datasheet is pretty clear. Switchable graphics is an option on T series (integrated only is the other option) and standard on W series.
No, the datasheet isn't clear. In fact they don't even mention that option in the W500 datasheet... I don't believe there will be Switchable graphics in W500 (unfortunately).
Posted: Wed Jul 23, 2008 5:36 pm
by hanbaoquan1518
There's really no point of arguing here, I guess we'll just have to wait and see.
Posted: Wed Jul 23, 2008 7:58 pm
by hanbaoquan1518
arlab wrote:awolfe63 wrote:I think the datasheet is pretty clear. Switchable graphics is an option on T series (integrated only is the other option) and standard on W series.
No, the datasheet isn't clear. In fact they don't even mention that option in the W500 datasheet... I don't believe there will be Switchable graphics in W500 (unfortunately).
And why is that???? Think about it, if you were Lenovo, what good would it be?
Posted: Wed Jul 23, 2008 8:58 pm
by tzcomwiz
so will the w500 be released on the same date as T400/500?
Posted: Wed Jul 23, 2008 9:00 pm
by awolfe63
Sounds like an opportunity for a wager. Loser buys winner a W500.

Posted: Wed Jul 23, 2008 10:00 pm
by Kel Ghu
Well, let's hope I am wrong... Especially that Nvidia has now annonced its new mobile GPU line.
http://www.dailytech.com/NVIDIA+GeForce ... e12434.htm
Posted: Thu Jul 24, 2008 12:38 am
by hanbaoquan1518
What's weird is that the W700 will also have the FireGL V5700, according to Thinkwiki. Which doesn't make any sense.
Posted: Thu Jul 24, 2008 8:42 am
by Pocket Aces
I guess we won't know until they're released! But otherwise, the W700 looks like the 17" version (my God, a 17" Thinkpad!) of the W500.
Posted: Thu Jul 24, 2008 10:42 am
by Marin85
I hope TP W700 will do even better than FireGL V5700

I mean there is a plenty of powerful graphic cards that would fit in a 17'' machine.
Posted: Thu Jul 24, 2008 1:24 pm
by Pascal_TTH
Don't trust too much GPU data base using 3D Mark 2006. CPU power can change the final score. Read the 3D Mark 2006 help file to understand how the final score is calculated.
I run 3D Mark 2006 yesterday on my T61p with any overclocking, just a clean Windows XP SP3 with leatest drivers :
Portable : Lenovo Thinkpad T61p
Processeur : Core 2 Duo T9300 à 2500 MHz
Processeur graphique : Quadro FX 570m 256 Mo DDR3 128 bits 475/700
Mémoire : 2 x 1 Go DDR2-667
Système d'exploitation : Windows XP
Driver : 174.71
Résolution : 1280x1024
Score 3D Mark 2006 : 4379
SM2.0 Score : 1737
SM3.0 Score : 1621
CPU Score : 2324
So it's much 13.3% higher than the 3862 score from notebookcheck.
Posted: Thu Jul 24, 2008 2:44 pm
by Marin85
I second that. Many people take the notebookcheck 3DMark 06 score table as a reference, but forget that there are also other factors that can affect the final score. Reading their score table one can often get a wrong impression about a given graphic card (e.g. that it performs bad although it could be actually a nice graphic card...)
Just my 2 cents
Marin
Posted: Thu Jul 24, 2008 3:55 pm
by hanbaoquan1518
Marin85 wrote:I hope TP W700 will do even better than FireGL V5700

I mean there is a plenty of powerful graphic cards that would fit in a 17'' machine.
I know, that's what it doesn't make any sense. At least an HD 3870 should be put in there.
Posted: Thu Jul 24, 2008 4:40 pm
by Pascal_TTH
Marin85 wrote:I second that. Many people take the notebookcheck 3DMark 06 score table as a reference, but forget that there are also other factors that can affect the final score.
Marin
Here is from the 3D Mark 06 help file :
3DMark06 Score Formula
The 3DMark Score is calculated using the following formula:
SM2.0 Score = 120 x 0.5 x (SM2 GT1 fps + SM2 GT2 fps)
HDR/SM3.0 Score = 100 x 0.5 x (SM3 GT1 fps + SM3 GT2 fps)
CPU Score = 2500 x Sqrt( CPU1 fps x CPU2 fps)
We'll define for clarity:
GS for hardware capable of running all graphics tests = 0.5 x (SM2S + HDRSM3S)
GS for hardware capable of running only SM2.0 graphics tests = 0.75 x SM2S
3DMark Score = 2.5 x 1.0/(( 1.7/GS + 0.3/CPU Score )/2)
i.e. 2.5 x weighted harmonic mean of GS and CPU Score.
GT1 fps means the average frame rate measured in SM2.0 graphics test 1. CPU fps means the frame rate measured in the CPU test.
Do an Excel sheet with those formulas and look how final score change according to CPU results.
Posted: Thu Jul 24, 2008 5:27 pm
by Marin85
Excuse me, but I don´t see much logic behind these formulas

I see they use 3 different average means (arithmetic, geometric and harmonic), though applied to the same unit, which makes things even more confusing because: if the same unit is used for testing different pieces of hardware, then there is quite much of an assumption that this unit follows structurally same relative change temps, which would be very suspicious if they haven´t justified this fact by using namely different estimators, which is now ok, but then namely because of this and the same unit (fps) for all hardware tests (for different pieces of hardware as well) putting the results in one expression is quite surprising!...

I think I will uninstall 3DMark 06... If they used some statistic data to adjust their formula for accuracy, i really doubt they would have come up with such simple estimators / average means... Anyway, I guess I´m too stupid to understand it right now...
Apart from that, I don´t trust the 3DMark06 score tables of notebookcheck because I could verify the differences by running 3DMark06 on my ThinkPad. The result here is much better than notebookcheck results and I´m really wondering how they tested ATI FireGL 5200 since all the notebooks on the market having this graphic card are AFAIK pretty equally equipped, so there shouldn´t be any big differences, but there are such...
Just my 3 cents
Marin
Posted: Thu Jul 24, 2008 6:25 pm
by Pascal_TTH
It' seems I hurt you sorry (I don't speak english very well). I agree with you. I just want to show how the score is build and thus why there is important differences in the final result for a same GPU with a small or a big CPU.

Posted: Fri Jul 25, 2008 5:31 am
by Marin85
Hey, buddy, don´t worry. It was nice of you to post the formulas

I had blind trust in 3DMark06 (at least regarding gaming performance) but this is not anymore the case. It was Futuremark that hurt me and probably I overreacted a bit

Anyway, thanks for the enlightenment
Marin
Posted: Fri Jul 25, 2008 2:08 pm
by Kel Ghu
Of course, 3DMark is a synthetic benchmark and like all benchmarks it has its flaws. Still, benchmarks are made to be indicative and 3DMark is a pretty good one, especially because now graphic cards have universal shaders. But here, I expect something like a 100% raw performance leap from one generation of T serie to the other. That means we always have got the middle-end card of the latest mobile graphic card generation, which has similar performance. That's what we had since T43p (maybe even before that). And any benchmark will show that the new card is substantially faster than the old ones. Anything below 30% (even 50%) performance increase is a detail and disappointing.
My point is, if V5700 is similar to HD3650 then we clearly will not have that 100% performance increase. Txxp series usually get middle-end card, right? Nvidia
slowest last generation card is the 9700M GT.
3DMark06
HD3650: 3450
Fx570: 3650
OCed FX570: 5600
9700M GT: 6250
Aren't you guys somewhat disappointed not to get even the low-end card of the latest graphic card generation?
Looks like people on notebookreview.com agrees with me.
http://forum.notebookreview.com/showthr ... ?p=3634557
Posted: Fri Jul 25, 2008 2:19 pm
by hanbaoquan1518
Where did you get the information that the 9700mGT is the low-end card??? It's a higher-end card compared to the Quadro 570m (8600GT GDDR3). That last generation of that card is the 8700mGT. You're comparing the last year GPU with this year GPU (a higher-end) one, what's the point???
Of course we won't have a 100% increase in performance, but it's certainly not worse.
Also, I"m the guy who posted in that thread too

Posted: Fri Jul 25, 2008 2:42 pm
by Icefang
I have a T61P and from everything I have seen so far, it doesnt seem to me that the new W500 will be a step forward in terms of graphic performace, it seems to be a step backwards actually. Last year the T61P had about the best graphics card you could find in a 15 inch notebook. This year their new model seems to regress while some other brands like Clevo / Sager have vastly improved their graphics chips.
http://www.powernotebooks.com/specs/Sag ... ecsxrp.php
I use my notebooks for both work and play. I boot them in linux for work and windows for games. Before buying a T60P 3 years ago I always used Sager's cause it was the only laptop I could get a decent graphics card in. I really like the Thinkpads, and appreciate the build quality / support difference they have over the Sager's but it seems to me the only way I am going to get a faster graphical machine in the next year, is to move away from the Thinkpads. I would much rather stay with the Thinkpad, but at best the W500 seems to be a push with the T61P when it comes to graphics capabilities. Been looking for benchmarks of the new Sager and the W500 but havent been able to find any as of yet. Hopefully some reviews start coming out soon.
Posted: Fri Jul 25, 2008 2:55 pm
by hanbaoquan1518
Clevo design their machines for gaming, and that's 15.4'' notebook is actually an oversize notebook.
The T61p or W500 is a balance between portability and power, not just power.
The W500 is not out yet, so we can't conclude anything about it. HP and Dell are also shipping their notebooks with ATI this time. So let's hope for something good.
Posted: Fri Jul 25, 2008 3:03 pm
by Pascal_TTH
GeForce 9700M GT was launch two or three weeks ago and NVIDIA gives scores by running 3D Mark 06 with a Core 2 Quad CPU. This GPU have the same number of SP than the Quado FX 570m. It's about the same with higher clock and new manufacturing process (60 nm ?).
http://portables4gamers.com/wp-content/ ... mark06.png
Core 2 Duo T9300 reachs 2300 for 3D Mark 06 cpu score. Look at the table : 3500 for cpu ! This is a Quad Core value ! So use the formula to compute the score with a common CPU laptop. You get about 5800~5900 with T9300.
I reach 5790 with Quadro @ 700/1400/999 with T9300 :
http://img329.imageshack.us/img329/3102/image5yr4.png
Posted: Fri Jul 25, 2008 3:12 pm
by Pascal_TTH
hanbaoquan1518 wrote:Clevo design their machines for gaming, and that's 15.4'' notebook is actually an oversize notebook.
The T61p or W500 is a balance between portability and power, not just power.
The W500 is not out yet, so we can't conclude anything about it. HP and Dell are also shipping their notebooks with ATI this time. So let's hope for something good.
They also switch because doing business with nvidia is a pity. Most manufacturers are fed up to work with them.
Posted: Fri Jul 25, 2008 4:27 pm
by Kel Ghu
Alienware has made a good gaming 15"4 too. Obviously, I wouldn't be here if I only cared of power. I need my laptop for CAD in chemical engineering and also need it to be rugged for travelling. CAD apps often need more CPU power than GPU and most people don't need FireGL for CAD. But those who plays are often those who appreciate the best graphic cards most imho. And I like to play when I am on the go. That's why I love Lenovo and that's also why I care this much about the W500. Bah, maybe i am better off buying a macbook pro
To hanbaoquan1518
That was just to add a little spice to the discussion. I felt people didn't care that much so...

Following Lenovo philosophy, W500 should get something similar to 9600 GT. Which is bad too, because it has similar performance as FX570.
Expect a new mobile line from AMD shortly, probably the 4000 serie and also probably something that is not a big improvement over 3000 serie. Just like nvidia 9000 serie.
I am comparing different generation of cards to show how bad is what people thinks is an upgrade. Doesn't it make sense?!?
Certainly not worse performance? That remains to be seen imho. Though that's not my point. My point is, it's a card that has very similar performance as FX570 and Lenovo should have chosen a much better card. But I guess mobile graphic cards have reach a thermal and power efficiency limit that hold back in the performance domain, that only physical improvements on GPU can change. It's been two generation now that don't have massive improvements on Nvidia 8000 line. Things are slowing down...
But yeah, you're right... Let's wait and hope for the best...
To Pascal
Oh yeah, my bad. 9700M GT sucks, similar card as FX570. We should be having 8800GTS or 9800GTS in W500 to keep up with the nearly 100% perf increase we have had in the past. Merci mec!
Being fed up with Nvidia is not a good reason to change graphic card supplier unless it doesnt hurt business, which is probably the case here. Nvidia has been so slow at improving their cards that AMD cards are now on par with Nvidias.
Posted: Fri Jul 25, 2008 6:12 pm
by arlab
hanbaoquan1518 wrote:arlab wrote:No, the datasheet isn't clear. In fact they don't even mention that option in the W500 datasheet... I don't believe there will be Switchable graphics in W500 (unfortunately).
And why is that???? Think about it, if you were Lenovo, what good would it be?
If I was Lenovo, I would state it clearly...
Anyway, I stand corrected. No Switchable graphics in W500.
Posted: Sat Jul 26, 2008 12:07 am
by hanbaoquan1518
I guess we have to wait, time will answer it all.

Posted: Mon Jul 28, 2008 1:16 pm
by Pascal_TTH
Lenovo is switching to ATI because NVIDIA GPU do not have displayport support buildin. In performances point, it's not a step foward.
Posted: Mon Jul 28, 2008 2:17 pm
by MadCap
Ok ppl, some real information here -
I got a sample ThinkPad W500

- W500 is equiped with switchable graphics > you can choose with option from power manager for high performance (V5700) or energy saving (X4500)
- the GPU is V5700 aka. HD3650, acutally this sample doesnt got final drivers, so its marked as HD3650 with 600 MHz on GPU and 700 MHz on memory (1 400 efectively)
- the gpu performance is little higher then 8600M GT on stock clocks (Pascal_TTH quadro is REALLy high clocked), more specific numbers later
- gpu switching helps the battery life a lot, mor specific numbers later
Any questions?

I will be returning it in a day or two ..
Btw > ati vs. nvidia > the new and more powerfull nvidia gpus are VERY new and this notebook is in development for some time for sure and the displayport is the cause too. If you want to compare performance - use reale applications. 3Dmark06 overall score is heavy cpu dependant and of course driver releated. Benchmark in games or in your cad/cam/professional apps.
The growth in performance in W500 vs. T61p is definitely low, or not breath taking, but there is lot more better and W510 (HD4xxx series gfx etc.) could be near, who knows

Posted: Mon Jul 28, 2008 4:09 pm
by Pascal_TTH
Hi MadCap,
Can you please run some benchs ? Even 3D Mark 06 will be nice if you can give the score with detail. Is the W500 noisy ? Does it heat ? Witch CPU have your W500 ? What about the hard drive ?
I overclock my Quadro for some benchs, just want to know how fast it can run. Default freq for performance 3D : 475/950/700 MHz (1400 effective). Regular 3D Mark 06 score with T9300 for my T61p is about 4350 ~ 4400 (higher than notebookcheck due to higher CPU).
It looks FireGL V5700 runs like regular Mobility Radeon HD 3650. So, it will offer less 3D power and it will also suffer from the same issue than the desktop Radeon 3650 : poor antialiasing and anisotropic performances.
Nice to know W500 have dual graphic engine ! This is the nicest feature of Centrino 2. X4500 is already a powerfull IGP.
Btw, thank you for the feedback !

Posted: Mon Jul 28, 2008 4:24 pm
by MadCap
Reviewing of notebooks is my job, so benchmarks/games/performance test are standard. I just don`t got it summarized and post it tommorow in clean and understanble form

.
I`am not going to overclock it - no time and I don`t support this on notebooks even while I like to overclock desktops

.
Noise and heat - very silent and cold. The cooling system is effective and the switching graphics helps it too - the power consumption goes down by 5W in standard usage when on Intel GFX.
Only complain so fra is the hard drive, which is little noisier then I like, but really fast - the new 2GB turbo module helps it a lot (got som numbers about its acceleration too).
P.S.>
my standard performance test contain this >
http://www.extranotebook.cz/files/image ... yvykon.png
and this
http://www.extranotebook.cz/files/image ... pvykon.png
the last four lines are - music compression, file compression, dvd conversion and avchd conversion