No it doesn't because the 5870m is already 128 bit with 1GB memory there would be no point lowering it further. Makes the 5870 even more fantastic by the power it throws out with the bandwidth, bus and memory it uses![]()
![]()
-
-
I didn't have time to get something to run benchmarks last night. Do any of you recommend a program to use?
-
The point is that GTX460m with 192bits memory bandwidth was slightly slower than Ati 5870m with 128bit memory.
The 128 bit 460m should be even slower.
Also please stop claiming that more memory doesn't matter (which is correct for lower resolution with no AA gameplay).
We're debating the performance penalty incurred (or not) by the lost memory bandwidth
Anyway: releasing a (to be proven) crippled version of the card under the same name and the same performance tag is kinda low and it reminds me of OCZ and their crippled 25nm SSD drives -
Even at 1920x1200P on my external screen with every feature active I have not used 1GB of memory usage. It may be possible in some instances but it is not a reason to change your entire outlook on what you buy.
However, the difference between 192bit and 128bit is marginal and to make a card slower than its competitor when it was already slower does not make sense unless your the person pocketing the money for it.
-
I would do 06 and 11 if possible, then we can compare them to already published #s. -
Ok. I'll run that as soon as I get home.
I know it's probably an ignorant question, but running the benchmark test doesn't put any degrading stress on the pc does it?
Also, when I ran GPU-Z... I had warcraft running in the background. I didn't figure this would effect the readouts but I was just wondering if maybe i should have done it when the computer was completely idle? -
Vantage GPU score might be a good indicator of whether or not the memory bandwidth and capacity impact performance compared to a 460m 1.5GB.
Be sure to run it at 1280x1024.
Edit: No, these benchmarks don't put any additional stress on the laptop than a game. Most people OC while benching to get higher numbers for bragging rights but for the sake of comparison please keep at stock clocks. -
The Bandwidth/Pixel Rate would change if under or over clocked but we can see by your GPU-z image this is not the case.
If you could test the above three MARK tests it would be very informative. -
Unfortunately I have the BST6 model which is only 900p which may be causing some discrepancies.
I'll try and run some of those benchmarks when I get home. Also, at this point I'm trying not to monkey with Over Clocking so it will definately be at stock clocks. -
-
Or you can run 3d Mark 11 in Performance P mode
-
So again, even if the buses were exactly the same, the 460 could have 8gb of vid ram and it will still perform almost exactly the same as a 1GB at ANY resolution. -
Doesnt matter how much memory you have, 512MB, 1GB or 1.5GB, you still have to use that memory and transfer things to/from memory.
Obviously that transfer is 50% faster on 192bits than on 128bits memory bandwidth.
How that translates in gaming performance, we're waiting on owners to post some scores...
PS. if the buses were indeed identical, 1GB or 1.5GB would not matter much -
1.5G will be useful for games with big texture files, like the High Res Textures for DA2. The extra RAM will allow for the hi res files + room for antialiasing
-
-
Sure, it will fit in memory, but run pretty much exactly how it would with 1gb. (For a 460. Of course for higher end GPU's 1gb vs 1.5 could help tremendously but not a mobility 460) -
Don't buy the BB model to avoid the " watered down effect".. Problem solved...
-
-
I don't think any model has been watered down that much though. Cutting back on the GPU in a gaming laptop? Not a good move Asus. I don't mind the lower resolution, slower hdd and less ram in my best buy model because I still have the most important part of the g73jh the GPU and CPU. I also like the warranty as it is the only reason I have my g73 and it will give me a new laptop in the future.
-
If we get a few bench marks to really see the difference then you would know. To honest, I'd look around more before I jump on a certain laptop. Yeah, the second gen i5/7 are nice, but you can still get some really crazy deals on the first gen with a better GPU. (5870) Ok, for $1100 plus taxes, which will run you $77 in Georgia, you get a i5 2410m with 4GB of RAM and a gimped 460m along with the 1366x768 res screen. I would say the GPU and screen are its bad faults. Now take look at Newegg, they have the MSI GX660 260 model, i5 460m, 4gb of RAM and 5870 gpu and 1366x768 screen. The CPU and screen are its bad faults. Same site, Asus G53JW-XN1 with the i7-740qm, 4GB of RAM, 1.5GB 460m with the 1080p screen for 1200 without tax unless you live in Cally. I want a new second gen Sandy bridge chip also and I have been eye balling every site out to see what coming out every say. I would say wait before you choose, because with in a month they will have out a lot more laptops with the new chips and who's not to say that Newegg or come other site wouldnt have a G53SW XA1/Xn1 or whatever they are going to call it with a 1080p screen and a true 460m for $1200. I've been stalking the websites trying to find the best deal for my money. I'm having a really hard time choosing what will be right for me because I like Asus, MSI and Sager at the moment. I'm just waiting to see what this new 560m is going to be really working with before I choose one.
-
Here's my 3d Mark 11 Benchmark for comparison now if anyone have the 1GB 460M GTX model then we could compare how's the performance.
-
-
-
-
The main reason for the VRAM difference is the screen resolution. The BB version res is 1400x900, thus 1gb does just fine, whereas the 1920x1080 can use the extra .5 gb memory for the increased res.
-
-
Hey,
Thanks,
-NRUserII -
If it is 1.5GB it will have the 192bit bus width as 128bit couldn't use or have any need of 1.5GB.
-
I got pissed after entering this topic, but now that I ran 30d mark i feel more relived... The difference in the graphics card is there but is not as big as i thought it would be considering the bandwidth bus...
Result
But i would like to see a benchmark on real games, that unlike that demos could be fillrate limited, or demand more bandwidth due to AA or whatever XD
My monitor is 1600*900 and i've been able to max every game at maxing settings at my native res so far, except for the Crysis games... I still can max them but i have to bring the res down to 720p (They still look gorgeous at that resolution, specially Crysis 2, so no complaints there).
In Crysis 2 everything being maxed out gives 30-40ish fps (never saw a dip below 30) with some areas going even higher. I limited the framerate to 30fps because that fluctuation all the time was really distracting. And even at 720 the game manages to look really sharp, and almost jaggie free due Crytek's custom AA. The IQ in this game completely smokes the one in Crysis 1. -
I posted early in this thread.
There is NO single memory controller in the gtx460m
gtx460m has three 64bits memory controllers. To reach 192 bits of memory bandwidth you have to populate with identical amounts of memory all 3 controllers.
That means total amounts of memory of 1.5GB or 768GB or anything (3GB like in lamborghini vx7?) that is divisible by 3
1024MB versions can only run over 2 memory controller and 2x64bits --> 128 bits)
For example GTX485 has four 64bits memory controllers and a 256bits wide memory interface while using 2GB of ram (or any amount of ram that is divisible to 4)
PS. Any 3D Mark 11 or 3d Vantage scores yet?
Edit:
@LukasTaves, well, it's slower... my firstgen i5 scores slightly better than your secondgen i7. And at higher ress and or better AA level, probably the difference will be bigger.
Pitty
Normal scores for 192bits gtx460 with i7-2630 are around 1830 with .658 drivers -
Any ideas on the difference in price to make up for this loss? -
Just Another thing.
The 1GB versions mite have lost not only some memory bandwidth but some of the ROP's as well (hopefully not)
I could not find any references to the 1GB version specifications but check the differences between gt 445m with 1GB and 1.5GB versions
Code:GT 445m (which is using the same gf106 as GTX460m): Fermi code name | memory, bandwidth | Shaders:TMU:ROP's gf 106 | 1024MB, 128bits | 144:24:16 gf 106 | 1536MB, 192bits | 144:24:24
-
-
This is ridiculous. They should have changed the name of the GPU since it isn't the same freakin' card. The should have changed the name of laptop as well from SW to something else.
-
-
On a more serious note, it's soleily Asus faults, not NVidia's -
The best example is the GF116 (560m), which is an optimized GF106 (460m) core with the same features all derivated from GF100 (480m). Clear? -
I'm not trying to start an argument, but better we do not point fingers since Ati did/does the same (an early example was radeon 8500 / radeon 9100 which was a blunt rename, a more recent example is - if i'm not mistaken - 5870m and 6870m) Clear?
Regarding the issue at hand, it is ASUS that makes the G73SW and when you buy that you expect a GPU with certain performances not a crippled version of it.
It should not bear the SW name but something else - SY/SZ/whatever -
A downclocked 5870m with a bigger price tag? Seriously ATI shame on you. -
G73SW-BB Limited Edition
Not the good Limited Edition, instead it is literally limited in features edition. -
Guys, if you notice, the retailer get's a personal sku (G73SW-XA1 for Amazon.com as an example) It's the retailer who configs what system and parts they want, Asus fulfills their order and suggests an MSRP, and the retailer sells it. If the system is not config'd for your tastes, then get another configuration from another retailer. The -A1 or A2 models are Asus configed, and aren't watered down. So don't go blaming Asus, blame the retailer for requesting 1G GTX460m's.
And you can blame them by not buying from them. -
-
-
-
Hi all.
I bought BB version with 900 res. three days ago and I have decide to return it. I have been running 3dmark06 for last two days and I wasn't getting those 13000..14000 scores like others do.. I was getting 8666 in 1024resolution and 9200 in 720resolution. I couldn't believe my eyes with thore ridiculous scores I was getting.. So I used driversweeper and installed the latest nvidia driver. I was getting same result but idle temperature for the CPU and GPU was about 10c higher than with oem driver. So, I went back to oem driver to reduce the temperatures and sure enough it ran 10c lower..
And I got so curious and saw this thread regarding gtx460m 1gb vs 1.5gb..
I found out why my BB version was getting 8000s in 3dmark06 for last three tries.. It is indeed Waterdown..Limited..Handicaped version whatever you call it. I just want people to know that BB version is not for gaming.. Stay away at all cost if you are serious gamer !! As for me I will be returning it tomorrow. My old Gateway with 9800m GTS was getting higher score than this.. Shame on BB..
Hope this helps..Attached Files:
-
-
-
Mite be something wierd with 3D Mark 06.
For example, i was never able to run it on my G53. I was getting EAL error while starting 3DMark 06. After solving it by putting the EAL dll in the right places, 3DMark06 kept crashing at start.
3D Mark 11 scores are bad but not to the same extent, although that could be due to the low ress used to generate P scores in 3DMark11 (720p)
Can anyone run 3D Mark Vantage in Performance mode with an external monitor? -
Just to be clear. The 5870 1GB cards in ALL the G73's (Best Buy or not) were indeed all the same right?
-
I think the 460m situation is a first... -
I feel sorry for those of us that dont know how much a difference the 1gb against 1.5gb card makes because a couple of years ago I didnt know anything about bandwidth of memory bus.
Lets just hope they look here and see this post.
GTX 460M (VRAM 1Gb vs 1.5Gb)
Discussion in 'ASUS Gaming Notebook Forum' started by NorthArmy, Mar 20, 2011.