ATI's New High End and Mid Range: Radeon X1950 XTX & X1900 XT 256MB
by Derek Wilson on August 23, 2006 9:52 AM EST- Posted in
- GPUs
What is GDDR4?
The major advancement that makes the new X1950 series possible is the availability of GDDR4. This really is an incremental step forward from GDDR3 with an eye towards power saving. Of course, in the high end graphics world "power saving" is a euphemism for overclockability. Thus we have a technology designed to run efficiently and to be pushed beyond the limit of reason. Sometimes we can have our cake and eat it too. While the majority of the power savings come in at lower clock speeds, we will see in our tests that there are some power benefits at the high end as well.
We have gotten our hands on some information about GDDR4, and will do our best to extract the most useful data. The major advances of GDDR4 include a lower voltage requirement of 1.5V (or up to 1.9V if overclocking). At the low end, this offers a 30% power savings over GDDR3 clock for clock. We also see a fixed burst length of 8 bits with GDDR4 as opposed to 4 with GDDR3. This allows the RAM to run at half the core frequency while offering the same memory bandwidth as GDDR3, which results in significant power savings (a 2GHz data rate GDDR3 chip would run with a core clock of 500MHz, while GDDR4 can run at 250MHz). Alternately, this can be used to provide higher memory speeds in high end systems.
Data bus inversion (DBI) also makes its way into memory with GDDR4. This technique helps to lower the average power used by the bus by minimizing the number of zeros transmitted. At first glance, this might not make much sense, but it all has to do with how zeros are sent. These days, it's most common to see digital logic use active low signaling. This means that a digital 1 is actually represented by a low power state. This is ostensibly because it is easier to create a sink than a source (it's easier to pull voltage down from a high state than to raise it up from a ground state). This means that we are actually using more power when we are sending a zero because the signal for a zero is a high voltage state.
The way DBI works is that all the data is inverted if the current byte to be transmitted contains more than 4 zeros. A separate control bit (aptly named the DBI flag) is used to indicate whether the data is inverted on the bus or not. Here are a couple examples of what would happen when transmitting data over a bus using DBI.
data to send: 11100000
data on bus: 00011111, DBI Flag = 1data to send: 11111000
data on bus: 11111000, DBI Flag = 0
Addressing is also done differently with GDDR4. If we are considering the 16Mx32 (this means 16 million address that hold 32bits of data each) 512Mbit GDDR4 modules currently available from Samsung, we will have only 12 address pins. A full address is sent in two consecutive clock cycles (as 24-bits are needed to select between 16 million addresses). This frees pins to use for other things, like power and ground which could increase the capability of the DRAM to run at high speeds. Among the other optimizations, a multi-cycle preamble is used to make sure that timing is accurate when sending and receiving data (allowing for faster speeds), GDDR4 has a lower input capacitance than GDDR3, and memory manufacturers have more control over the properties of the transistors and resistors used in the driver and receiver in order to better tune products to specific needs.
Right now, ATI is using Samsung's 80nm 0.91ns K4U52324QE GDDR4 modules on its X1950 products. This is actually the slowest GDDR4 memory that Samsung sells, clocking in at a max of 1100MHz. Their 0.714ns RAM is capable of hitting 1400MHz which will be able to put future graphics cards beyond the 2.5GHz data rate and up near the 80GB/s range in memory bandwidth. Of course, the X1950 XTX memory bandwidth of 59.6GB/s is pretty impressive in itself. From a clock for clock perspective, GDDR4 can offer advantages, but we shouldn't expect anything revolutionary at this point. We ran a couple tests underclocking the X1950 XTX, and saw performance on par with or slightly faster than the X1900 XTX.
Facebook
Twitter
RSS
74 Comments
View All Comments
JarredWalton - Wednesday, August 23, 2006 - link
We used factory overclocked 7900 GT cards that are widely available. These are basically guaranteed overclocks for about $20 more. There are no factory overclocked ATI cards around, but realistically don't expect overclocking to get more than 5% more performance on ATI hardware.The X1900 XTX is clocked at 650 MHz, which isn't much higher than the 625 MHz of the XT cards. Given that ATI just released a lower power card but kept the clock speed at 650 MHz, it's pretty clear that there GPUs are close to topped out. The RAM might have a bit more headroom, but memory bandwidth already appears to be less of a concern, as the X1950 isn't tremendously faster than the X1900.
yyrkoon - Wednesday, August 23, 2006 - link
I think its obvious why ATI is selling thier cards for less now, and that reason is alot of 'tech savy' users, are waiting for Direct3D 10 to be released, and want to buy a capable card. This is probably to try an entice some people into buying technology that will be 'obsolete', when Direct3D 10 is released.Supposedly Vista will ship with Directx 9L, and Directx 10 (Direct3D 10), but I've also read to the contrary, and that Direct3D 10 wont be released until after Vista ships (sometime). Personally, I couldnt think of a better time to buy hardware, but alot of people think that waiting, and just paying through the nose for a Video card later, is going to save them money. *shrug*
Broken - Wednesday, August 23, 2006 - link
In this review, the test bed was an Intel D975XBX (LGA-775). I thought this was an ATI Crossfire only board and could not run two Nvidia cards in SLI. Are there hacked drivers that allow this, and if so, is there any penalty? Also, I see that this board is dual 8x pci-e and not dual 16x... at high resolutions, could this be a limiting factor, or is that not for another year?DerekWilson - Wednesday, August 23, 2006 - link
Sorry about the confusion there. We actually used an nForce4 Intel x16 board for the NVIDIA SLI tests. Unfortunately, it is still not possible to run SLI on an Intel motherboard. Our test section has been updated with the appropriate information.Thanks for pointing this out.
Derek Wilson
ElFenix - Wednesday, August 23, 2006 - link
as we all should know by now, Nvidia's default driver quality setting is lower than ATi's, and makes a significant difference in the framerate when you use the driver settings to match the quality settings. your "The Test" page does not indicate that you changed the driver quality settings to match.DerekWilson - Wednesday, August 23, 2006 - link
Drivers were run with default quality settings.Default driver settings between ATI and NVIDIA are generally comparable from an image quality stand point unless shimmering or banding is noticed due to trilinear/anisotropic optimizations. None of the games we tested displayed any such issues during our testing.
At the same time, during our Quad SLI followup we would like to include a series of tests run at the highest possible quality settings for both ATI and NVIDIA -- which would put ATI ahead of NVIDIA in terms of Anisotropic filtering or in chuck patch cases and NVIDIA ahead of ATI in terms of adaptive/transparency AA (which is actually degraded by their gamma correction).
If you have any suggestions on different settings to compare, we are more than willing to run some tests and see what happens.
Thanks,
Derek Wilson
ElFenix - Wednesday, August 23, 2006 - link
could you run each card with the quality slider turned all the way up, please? i believe that the the default setting for ATi, and the 'High Quality' setting for nvidia. someone correct me if i'm wrong.thanks!
michael
yyrkoon - Wednesday, August 23, 2006 - link
I think as long as all settings from both offerings are as close as possible per benchmark, there is no real gripe.Although, some people seem to think it nessisary to run AA as high resolutions (1600x1200 +), but I'm not one of them. Its very hard for me to notice jaggies even at 1440x900, especially when concentrating on the game, instead of standing still, and looking with a magnifying glass for jaggies . . .
mostlyprudent - Wednesday, August 23, 2006 - link
When are we going to see a good number of Core 2 Duo motherboards that support Crossfire? The fact that AT is using an Intel made board rather than a "true enthusiast" board says something about the current state of Core 2 Duo motherboards.DerekWilson - Wednesday, August 23, 2006 - link
Intel's boards are actually very good. The only reason we haven't been using them in our tests (aside from a lack of SLI support) is that we have not been recommending Intel processors for the past couple years. Core 2 Duo makes Intel CPUs worth having, and you definitely won't go wrong with a good Intel motherboard.