backtop

Print E-mail del.icio.us 13 comment(s) - last by wordsworm.. on Dec 12 at 7:28 AM

NVIDIA GeForce 8800 GTS 512MB @ AnandTech
DailyTech's roundup of hardware reviews from around the web for Tuesday

NVIDIA GeForce 8800 GTS 512MB
@ AnandTech
@ Driver Heaven
@ PC Perspective
@ Hardware Canucks
@ Hot Hardware
@ Dionx PC
@ bit-tech
@ Legit Reviews
@ FPS Labs
@ I4U

Notebooks
Dell XPS M1730 Gaming Notebook @ HotHardware.com

Memory
Patriot Extreme Performance 2GB PC2-9600 DDR2-1200MHz @ OCC

Video
ATI RADEON 3850/3870 (RV670) @ Digit-Life
Leadtek WinFast 8500 & 8600GT with HDMI @ TweakTown
NVIDIA GeForce 8800 GT @ PenStar Systems

Cases
Antec Sonata III 500 @  BurnOutPC.com
Sunbeamtech Quarterback @ Hi-Techreviews.com

Cooling
Nanoxia Dualcontact Hybrid Base CPU Cooler @ Tweaknews.net
Scythe Ninja Mini SCMNJ-1000 Heatsink @ Frostytech.com
Glacialtech Igloo 5610 Silent Heatsink @ Frostytech.com
Gigabyte 3D Galaxy II Water Cooling System @ OCModShop

Storage
OCZ ATV 2GB & ATV Turbo 4GB @ Techgage
Rosewill RX81-CW-US Hard Drive Enclosure @ ThinkComputers.org

Digital Cameras
Canon PowerShot G9 @ Trusted Reviews

Comments     Threshold


This article is over a month old, voting and posting comments is disabled
Anandtech's article
By wordsworm on 12/11/2007 10:41:15 AM , Rating: 2
I just read Anandtech's article comparing the various flavors of 8800s out there and I have a question someone might be able to answer. The article stated that they're using 4GB of motherboard memory and videocards ranging in memory from 256MB to 768MB. If 32bit OS can only address 4GB, wouldn't the results of the tests be influenced and therefore flawed? If you use 256MB of videocard memory, then you have 3.7GB of motherboard memory, whereas for the GTX you have only 3.2GB of motherboard memory.

If that's not how it works, please, someone enlighten me.




RE: Anandtech's article
By jtemplin on 12/11/2007 10:50:29 AM , Rating: 2
I posted something about this topic some time ago... I think you have a point here, but I think it would only matter if this limit is being reached(3.2). I feel like when they run a benchmark they most likely have ever unnecessary process/service killed. So my gut tells me that what your saying is a problem but only if any test exceeded 3.2Gb of usage, at which point there would be paging, and probably a massive frame-rate hit. Thats my .02


RE: Anandtech's article
By Pryde on 12/12/2007 12:21:18 AM , Rating: 2
A single program can only address 2GB of space. Even if you lose 1GB due to how its managed then you still have 1GB for programs in the background.

So as long as the background programs are using less than 1GB of ram then the game still has access to the max 2GB and won't have any affect on performance.


RE: Anandtech's article
By Curelom on 12/11/2007 10:51:50 AM , Rating: 1
The answer is no. The video card memory is not managed by the OS, but by the video card firmware, so you can run 4GB of memory on the mobo and however much plus on the video card.


RE: Anandtech's article
By DkFFIV on 12/11/2007 11:25:00 AM , Rating: 2
Unless you're running a 64-bit OS the computer will not be able to address all 4GB of memory. I have a 768MB 8800 GTX with 4GB of RAM and XP only recognizes 3.25GB of memory.


RE: Anandtech's article
By Etsp on 12/11/2007 12:22:31 PM , Rating: 3
You could have a 64MB video card and still be only able to access 3.25GB of your ram... that's the OS's default limit...


RE: Anandtech's article
By kkwst2 on 12/11/2007 2:09:31 PM , Rating: 2
Why the heck was this rated down? This is correct. The amount of memory your video card has is irrelevant. It's not addressed directly by the operating system or the motherboard. The video card manages how this memory is used directly.


RE: Anandtech's article
By wordsworm on 12/12/2007 7:28:21 AM , Rating: 1
I think it was rated down because the OS is important for allocating hardware resources for software. From all the research I've done so far, I cannot help but conclude that the reviews done with 4GB of memory on 32bit OSes will have skewed results because the amount of usable RAM will depend on the amount of videocard ram used. There's a few ways to find out for sure, and that's to do all of these tests with 2GB and 64 bit Vista. I'm willing to bet that there would be a 1-2% gap decrease between the GTX and 512MB videocard, and even 3-4% difference increase between the GTX and the 256MB version just because the way the memory would be handled. Of course, I have no way to prove my statement, much less the resources to prove it.

Nonetheless, I think my theory is likely fact, that any results posted using 4GB of memory with a 32 bit OS can cloud 'precise' figures. According to people out there, differences in RAM on the video card does affect the amount of RAM that shows. But please, do your own research, don't take my word for it. I don't have a personal interest in this debate other than the desire to figure out whether these numbers provided by the tech rags out there can be counted on as precise and reliable.


RE: Anandtech's article
By wordsworm on 12/11/2007 4:56:00 PM , Rating: 2
I found a number of articles addressing this:
http://www.upgrade2vista.com/index.php/33/maximum-...
http://www.vistaclues.com/reader-question-maximum-...

Both of these state that the size of the RAM on the video card is an issue. Would you mind providing me with a citation to something that suggests otherwise?

Someone else mentioned that Windows 2000 allowed more than 4 GB of memory, but this article here: http://www.brianmadden.com/content/article/The-4GB...
mentions that
quote:
People who are unfamiliar with the real meaning behind the 4GB Windows memory limit often point out that certain versions of Windows (such as Enterprise or Datacenter editions) can actually support more than 4GB of physical memory. However, adding more than 4GB of physical memory to a server still doesn't change the fact that it's a 32-bit processor accessing a 32-bit memory space. Even when more than 4GB of memory is present, each process still has the normal 2GB virtual address space, and the kernel address space is still 2GB, just as on a normal non-PAE system.


According to all of the articles I quoted, 32 bit means a maximum of 4 physical GB for both the motherboard and the video card. The extra RAM on the server edition is a workaround, and not a true 4+ GB solution. So, can you or someone else reference an article that explains that the video card memory isn't managed by the OS?


RE: Anandtech's article
By Screwballl on 12/11/2007 12:07:47 PM , Rating: 4
By definition, a 32-bit processor uses 32 bits to refer to the location of each byte of memory. 2^32 = 4.2 billion, which means a memory address that's 32 bits long can only refer to 4.2 billion unique locations (i.e. 4 GB).

Windows 2000 32-bit did have the option to address up to 64GB of memory... there is a special version of the core kernel image (Ntoskrnl.exe) with support for PAE called Ntkrnlpa.exe. (The multiprocessor version is called Ntkrpamp.exe.) To select this PAE-enabled kernel, you must boot with the /PAE switch in Boot.ini.

This special version of the kernel image is installed on all Windows 2000 systems, even Windows 2000 Professional systems with small memory. The reason for this is to facilitate testing. Because the PAE kernel presents 64-bit addresses to device drivers and other system code, booting /PAE even on a small memory system allows a device driver developer to test parts of their drivers with large addresses. The other relevant Boot.ini switch is /NOLOWMEM, which discards memory below 4 GB and relocates device drivers above this range, thus guaranteeing that these drivers will be presented with physical addresses greater than 32 bits.

Only Windows 2000 Advanced Server and Windows 2000 Datacenter Server are required to support more than 4 GB of physical memory.

So some people can switch back to Win2000 Pro and be allowed usage of all 4GB or more if they choose with these switches.


RE: Anandtech's article
By anotherdude on 12/11/2007 5:38:20 PM , Rating: 2
On a consumer 32 bit OS such as Vista or XP the OS can address only 4 gig total memory, for the reasons Screwball outlined. So if you have 4 gig of system RAM (I.E. 4 Gig installed in MOBO slots) a chunk of addresses must be reserved for other devices which have memory on them, most notably the video card which has a LOT of RAM on it. The more RAM the video card has then the more addresses the OS must reserve, theoretically. In practice those using the 8800 GTX with 700+ MB do seem to end up with less available RAM out of their 4 gig installed sticks than do others with less video RAM although the exact number does not always seem to be what you might expect from simply subtracting out the video ram.

This could cause the OS to run out of RAM faster when running the higher RAM video card and if this were to happen and this caused hard drive paging then the testing results could possibly be affected.

That's how I understood the question.


By RussianSensation on 12/12/2007 1:01:12 AM , Rating: 2
Not 1 game tested in the AnandTech review even benefits from having more than 3.25GB of system RAM. Whether the system has access to 3GB, 3.25GB, 3.7GB or full 4GB would have absolutely no impact on the outcome of those benchmarks. If the test bed used 2GB of ram, the outcome of the conclusions would have remained the same given the current games tested.




The rest
By wordsworm on 12/11/2007 8:48:32 PM , Rating: 1
I managed to glance at the rest of the systems and discovered that most of them don't even say which OS they're running: they just say Vista Ultimate. So, is it 32 bit or 64 bit? That is the question. At least Anandtech make this important little detail known. I assume that they're all running 32 bit software. Tonigh I'll have to calculate if there's a greater difference or an equal difference between systems tested with 2 GB and those with 4 GB of RAM between the various video cards.




"Nowadays you can buy a CPU cheaper than the CPU fan." -- Unnamed AMD executive









botimage
Copyright 2009 DailyTech LLC. - RSS Feed | Advertise | About Us | Ethics | FAQ | Terms, Conditions & Privacy Information | Kristopher Kubicki