The Gigabyte GA-X48T-DQ6 - Redefining the High End?
by Rajinder Gill on January 2, 2008 3:15 AM EST- Posted in
- Motherboards
Windows XP Benchmarking Results
Qx9650 Everest Cache and Memory Benchmark
9x333
 |
9x400
 |
9x450
 |
10x400
 |
One thing that stands out for us is that overall memory access latency is below the figures we have come to expect from DDR3 on the X38/X48 chipset. Although not shown here, this pattern also continues to the 65nm process CPUs. Apart from tRD and the finicky tRD phase shifts (see the BIOS guide later in this review), there are no other chipset registers we can access to easily increase 1N latency performance. We even tried an E6850 to see if increased levels of FSB would allow a linear scaling of memory access latency. Again, the figures were off the pace, coming in around 57ns at 500FSB (3.5GHz) with a tRD of 7. How this pans out into real world benchmark tests is always interesting, especially as most 3D game engines seem ambivalent to overall chipset performance.
Facebook
Twitter
RSS
37 Comments
View All Comments
Frumious1 - Thursday, January 3, 2008 - link
What are you, the jilted lover of Raja or something? Seriously, I can't say I blame the AT crew for taking a break. Me and my family certainly did our fair share of slacking off over the holidays.Merry Christmas, Foxy Scrooge....
8steve8 - Wednesday, January 2, 2008 - link
i just got 8gb od ddr2 800 1.8v for $180 shipped from newegg.8GB of DDR3 costs $1,150 shipped from newegg...
thats not at typo.
DDR3 is irrelevant for now.
10x cost for 3% perf boost...
I'm not even going to read this review.
Kougar - Wednesday, January 2, 2008 - link
This makes no sense. Gigabyte offers GTL controls on the P31-DS3L, values are: 0.636, 0.603, 0.566, 0.54. Why offer them on a $80 budget board but not a ~$300 X48T end-of-the-world board?? I was expecting them to be there when I heard about the redesign.This was a great review, and I am still reading/rereading it to absorb the info. I had a few questions though... You said FSB Overvoltage control requires 1.4v for both processors, and stock is 1.15v?? I never touched this setting by more than +1 with my Q6600, but it might explain a few things. How or where did you find out what the stock VTT value was, as I wish to be able to look up this info myself for future reference and check the P35 version.
Rajinder Gill - Wednesday, January 2, 2008 - link
I used a DMM and probed both the supply mosfet and GTL transistors.Regards
Raja
Kougar - Wednesday, January 2, 2008 - link
Thanks for your reply! Are there any guides or what do you recommend for novice multimeter users trying to discern which mainboard components power what. I don't mind doing the research, I just need a pointer at where to get started. I have tried before but was not sure if I was measuring the correct components.Just for proof, here is a screenie of the P31 DS3L $75 board's GTLREF menu: http://i111.photobucket.com/albums/n129/Chanur64/I...">http://i111.photobucket.com/albums/n129/Chanur64/I...
I can't fathom why the X48-DQ6 lacks this, could they be waiting to add it in later when the BIOS is more mature?
PLaYaHaTeD - Wednesday, January 2, 2008 - link
Since nobody cares to ask, I will go ahead:When can we buy an X48 Motherboard, along with a Q9450 cpu? How can you give such a comprehensive preview, and talk about recommendations without addressing a release date?
If some of the rumors are true about X48/Wolfdale/Yorkfield being delayed to possibly march, then wouldn't it be irrelevant to speak about recommending this product 2-3 months before it will even be available?
Gary Key - Friday, January 4, 2008 - link
ETA of the X48 boards is unknown. They are ready, from a BIOS viewpoint the Gigabyte boards need some more work but the ASUS boards that will be reviewed next week are about 97% there. Boards have been manufactured by all of the majors, it is up to Intel to pull the trigger. It was going to be in December, moved to 1/7/08, moved to 1/21/08, and is now in a holding pattern waiting on X38 stocks to clear out to some degree, which is surprising as most of the suppliers will move X38 downward to replace the mid-market P35 boards.The Q9450 and others will be announced shortly and official launch dates will be available at that time. Once again, up to Intel, but after the Phenom launch, they have no real reason to hurry up. The reason for the previews is provide just that, a preview, but we were expecting (along with Gigabyte/ASUS/MSI) that boards would be launched by the 21st of this month.
minimeat - Wednesday, January 2, 2008 - link
Good review on the BIOS features, i've been looking for this kind of detail for each timing number to be explained for a while, as none of this information is in the Gigabyte manual for any of their motherboards (HINT HINT).Anyways, you forgot to explain one really important number that i have been wondering about for a while, the Refresh to ACT Delay. Can someone please explain this number, either update the review or post it here, it would be much appreciated. I noticed that they had it set to 60 in the review, and there is no Auto option inherent in the motherboard's bios for this number, and i have no clue what it does. Thanks!
Glenn - Wednesday, January 2, 2008 - link
Looks like Anands crew spent plenty of time with this board! Amazing that no comparison benchmarks are shown against more mainstream boards like 965s, P35s or some AMD flavors? All that work with very little usefullness outside of early adoptee's!Gary Key - Friday, January 4, 2008 - link
Hi,We will have a Intel chipset comparison roundup the week of the 28th, maybe earlier if Intel decides to launch this product according to the last timeline. We will have the ASUS X48 boards up next week with a very detailed MCH overview and the MSI X48 board the week after with some Quad CrossFire loving if the drivers make it. These previews are concentrating on getting the most out of the board, the roundup will concentrate on comparisons to other available products.