DFI UT X48-T3RS LanParty - Living Review - Page 3

Clunk · #21 31-05-2008, 05:44

The Genie BIOS - Part 9

The Voltages - Continued.

CPU VTT Voltage range: - (1.1v - 1.518v) (These values will vary slightly between 45nm & 65nm CPUs)
This ties in closely with CPU GTL ref, so if you start tweaking this, you will need to keep an eye on the GTL REF as well - See the GTL REF/VTT Guide.

CPU VTT Voltage - Lower Half.

Clockgen Voltage Control: (3.45v - 3.85v)

This is the overall IC working voltage. You can increase this if you are going for insane FSBs, otherwise, keep it at 3.45v.

GTL+ buffer Strength: (Strong / Weak)

North-Bridge reference voltage strength adjustment.

Host Slew Rate: (Strong / Weak)

North-Bridge voltage driving strength adjustment.

GTL REF Voltage control: Enable / Disabled

"CPU VTT reference voltage for determining host bus high / low level." - This setting can be useful or detrimental and is highly dependant on your individual CPU. You may find that you can acheive a higher and more stable overclock by tweaking these values. The values are strongly tied in with the CPU VTT voltage, so it can take a bit of experimenting to find the correct settings. See the VTT/GTL REF guide here.

Enabling the GTL REF setting will allow you to fine tune the percentage figure should your setup require it, however, it is very easy to tweak these settings too much +/- and miss that sweet spot, and often, you will find that the sweet spot was at 67%/63% to begin with.

I have tested with a Q6600 G0 and this responded well to the 67% (Disabled) setting. Again, this doesn't automatically mean that yours will. My last Q6600 G0 responded better to fine tuned GTL REF and a slight bump in VTT, so it really is a case of experimenting and seeing what works best for you.

There is another factor that can have adverse effects on your VTT/GTL tweaking results and that is temperature. From my own tests, I have found that GTL REF tweaking is less effective when using air cooling and a higher Vcore, and similarly, when using sub zero cooling, I found the GTL REF to be incredibly sensitive and the slightest changes caused adverse affects.

Also, if you would like a more in depth read on GTL REF, click here to go to EdgeOfStability.com.

GTL REF Set To Enabled will show the 3 GTL Settings below.

Continued below...

Clunk · #22 31-05-2008, 05:44

The Genie BIOS - Part 10

Back To The Main Genie BIOS Page.

Exit Setup Shutdown: Mode1/Mode2

Intel boards are well known for shutting down unexpectedly after changing settings. DFI has given us two modes for this.

Mode 1 - If the FSB hasn't changed too much (value unknown at this time) the system will run a quick self diagnostic test and if everything is ok, it will skip the shutdown stage and rewrite the clock generator directly.

Mode 2 - Regardless of how much the FSB has changed, the system will shut down and reboot after you have changed any settings.

Shutdown after AC Loss: Enable/Disable If enabled, this will cause the system to automatically restart after a power failure.

OC Fail Retry Counter: 0~3 times

If the system fails to boot because of a bad overclock, this setting determines the amount of times that the system will attempt to restart before it reverts the FSB back to default to boot the system. If you set it to 1, it will attempt to restart once before resetting the FSB.

CPU Clock ratio: Exactly what it says - This is the Multi for your CPU. 6-11 for locked processors, 6-50 for unlocked processors.

CPU N/2 Ratio - This enables .5 multis.

CPU Clock: Enter your requred FSB here.

Continued below...

Clunk · #23 31-05-2008, 05:44

The Genie BIOS - Part 11

Back To The Main Genie BIOS Page.

Boot-up clock: Auto/ 100MHz ~410MHz

This setting acts as a buffer if you have your FSB set too high to boot successfully. If you set it to 100Mhz, it will drop the FSB by that value, until the system is booted, and then it will increase the FSB to the higher value.

Boot Up Clock - Lower Half.

CPU Clock Amplitude.

CPU Clock0 Skew - Top Half.

CPU Clock1 Skew - Top Half.

CPU Clock1 Skew - Lower Half.

Continued below...

Clunk · #24 31-05-2008, 05:45

The Genie BIOS - Part 12

Back To The Main Genie BIOS Page - Continued.

DRAM Speed: This is where you set your Memory Strap / Divider, call it what you will.

See also here DRAM Speed Converted To DRAM Ratios.

DRAM Speed - Lower Half.

PCI-E Clock (100-250) - Sets the speed for the PCI-E BUS. 100 is fine for even some brutal overclocking.

PCI-E Slot Config - This just sets the way the bandwidth is distributed between the PCI-E slots. For most systems, you can leave it at default.

CPU Spread Spectrum - For best results when overclocking, set to disabled.

PCI-E Spread Spectrum - For best results when overclocking, set to disabled.

That's all of the BIOS pics, next up, first tests...

Clunk · #25 31-05-2008, 05:45

First Tests & First problems [Fixed]

Problem 1 - [Fixed]

While conducting the first series of tests, I noticed that my PCI-E link width was at 1x and it should have been at 16x.

I had my 8800GT in the second PCI-E slot, which is supposed to be 16x. I double checked the documentation, and sure enough, it is supposed to be 16x.

There is a setting in the BIOS to change the PCI-E config, but on previous boards, X38/X48, I never needed to change this to have a single graphics card running at 16x in the second 16x slot.

I moved the card to the third PCI-E slot down, and it still showed 1x in CPU-Z.

The BIOS was set to 1x 1x and the only other setting was 4x, so I changed to that and I was able to get the card running at 4x in the second PCI-E slot down.

Update 06-06-2008: It turns out that there is an error in the user manual and PCI-E 4 is NOT a 16x slot. The 16x slots are 1 and 3. I had to reset the CMOS to get it to work, but it is all ok now.

Problem 2 - [Fixed/Explained]

Am am using an E8500 and the VID is 1.1v, however the BIOS detects the VID as 1.22v - Not a massive problem, but it needs fixing.

Anyway, those problems aside, the board is working fine so far.

Update 06-06-2008: DFI said "I have check the SPEC. @intel and award BIOS Inc. , they said the 45nm CPU boot with default voltage is 1.1V right, but Intel recommend /provide its CPU working Vcore table for disktop is 1.2V , so BIOS/Intel source code will select it on around 1.2V ."

From that, I suspect that this isn't in fact a problem, but is normal. Looking further down the page, it does show the VID as 1.1, but then 1.22v at the top.

First Tests.

Here we have a couple of quick tests using air cooling that I did to get the feel of things.

400x9.5 @ 1.36v - 3800Mhz CPU - 1600Mhz RAM @ 1.8v - 9-9-9-24

Click Images For Full Size!

410x9.5 @ 1.365v - 3895Mhz CPU - 1640Mhz RAM @ 1.8v - 9-9-9-24

Click Images For Full Size!

420x9.5 @ 1.4v - 3991Mhz CPU - 1680Mhz RAM @ 1.82v - 9-9-9-24

Click Images For Full Size!

More on the way...

Clunk · #26 31-05-2008, 05:45

DFI UT X48-T3RS Voltage Measuring Points.

For those who want to work out their own GTL REF charts, or just want to check if their voltages are correct, here are the common ones.

Warning: If you don't know what you are doing, please don't go poking round with a DMM, you will kill your board - I accept no responsibility for any damage or loss caused by using the info here.

Key

1. CPU Vcore Voltage (C535).

2. NB Core Voltage (LI12).

3. CPU GTL REF Voltage 1/3 (U61 - Pin 1).

4. CPU GTL REF Voltage 0/2 (U63 - Pin 1).

5. NB GTL REF Voltage (U52 Pin 1).

6. FSB VTT Voltage (D6).

7. DRAM Voltage (EC109 +).

Thanks to DFI for the info.

More tests on the way...

Clunk · #27 31-05-2008, 05:45

More Tests & Results With the Corsair DDR3-1600 Kit.

I'm still using the shipping BIOS as I have seen quite a few reports of people not being able to boot with their 2GB modules installed. The Corsair modules booted first time and the results have been quite impressive so far.

CAS 7 wasn't doable with these modules on this BIOS unfortunately, but I managed a fairly healthy CAS 8 @ 1600Mhz at stock voltage.

400x9.5 @ 1.30v - 3800Mhz CPU - 1600Mhz RAM @ 1.8v - 8-8-8-18 - 1:2

Cinebench.

430x9.5 @ 1.33v - 4086Mhz CPU - 1720Mhz RAM @ 1.82v - 9-9-9-24 - 1:2

Cinebench.

440x9.5 @ 1.36v - 4178Mhz CPU - 1760Mhz RAM @ 1.82v - 9-9-9-24 - 1:2

Cinebench.

At this point, I started to have some problems with OCCT, and I don't think it was due to stability issues with the system as prime95 worked fine.

The issue was that OCCT just hung, but didn't display a warning, and yet the rest of the PC carried on as normal. I haven't seen this before on my systems, so at the moment, I'm not sure what to make of it.

While trying to get OCCT stable, I relaxed the tRCD to see if it made any difference, and it didn't, however, I forgot to change it back for this quick prime run - I will redo it later.

450x9.5 @ 1.38v - 4276Mhz CPU - 1800Mhz RAM @ 1.82v - 9-10-9-24 - 1:2

Cinebench.

More results on the way...

Clunk · #28 31-05-2008, 05:45

First Tests With OCZ OCZ3P2000EB2GK 2x1GB OCZ PC3-16000 2000MHz Platinum EB Enhanced Bandwidth (9-8-8) Dual Channel kit.

As I had gone as far as I could with the Corsair 2x2GB kit, I decided to fit the niftily titled OCZ OCZ3P2000EB2GK kit and update the BIOS to the latest official Beta release 0609.

The yellow slots gave the best stability at higher speeds with the 2x2GB kit, so I'm going to stick with those for now.

460x9.5 @ 1.38v - 4276Mhz CPU - 1840Mhz RAM @ 1.82v - 9-8-8-24 - 1:2

470x9.5 @ 1.43v - 4466Mhz CPU - 1880Mhz RAM @ 1.82v - 9-8-8-24 - 1:2

480x9.5 @ 1.45v - 4561Mhz CPU - 1920Mhz RAM @ 1.82v - 9-8-8-24 - 1:2

I came to a grinding halt here using the 9.5 multi and the 1:2 strap. Still, it's a pretty respectable overclock.

488x9.5 @ 1.53v - 4561Mhz CPU - 1952Mhz RAM @ 1.82v - 9-9-9-24 - 1:2

More results on the way...

Clunk · #29 31-05-2008, 05:45

Some General Observations.

The shipping BIOS was fairly well behaved until you start overclocking heavily, and then a few restrictions appear when using 2GB modules.

There has been much discussion regarding 2GB modules from several manufacturers that fail to post in the T3RS, fortunately, my Corsair 2GB modules have performed well and only lost stability around 1820Mhz.

After updating to Beta BIOS 0609, I was slightly disappointed that I haven't been unable to hit 500FSB/2000Mhz with any kind of stability with the OCZ modules.

There have been reports of people managing this, but they all appear to have been using an unreleased Alpha BIOS. (I have the BIOS in question, but I feel that the results won't be relevant to the review if I use a BIOS that everyone else doesn't have access to.

Recovery From A Bad Overclock.

I've been saying this for a while, and I feel that this is an area of the Genie BIOS that needs looking at, and that is the board's ability to recover from a bad overclock.

Most of the time, the recovery is fine and it will simply restart and let you get on with it.

Quite often though, it will fail to post and a CMOS clear is required, and sometimes the board needs to be powered down for a while before it will post.

This scenario is generally caused by going too far with the memory, so...

In an ideal world, I would like to be able to press a key at start up, and it would reset the NB strap and maybe reset the main timings to default so you can get back to the BIOS without having to reset the CMOS each time.

Currently, you can reset the FSB at startup, but this doesn't work if your overclock failure is a result of going too far with the memory!

Of course, you can save your settings to CMOS reloaded, but it can be a pain if you forget!

Clunk · #30 31-05-2008, 05:45

Conclusion.

It’s been a slightly unusual review, in so much as there were a lot of results at the beginning and then not much to report since.

The Bad News.

After the excellent performance of the LT X48-T2R, it was slightly disappointing to find that the T3RS would not allow me to pass 488FSB stably, regardless of the multi, NB strap or indeed voltage.

Firstly, I suspected that my CPU may have degraded, so I fitted it into the T2R and it did 520x9.5 without any problems, so the CPU was fine.

All the memory that I was using had Samsung ICs and DFI said that the problem could be that those ICs like a tRFC of 100+, but the board only went up to 77. They issued me with several beta BIOSes, but I was unable to pass 488FSB.

During this time, Chris (Supershanks) also had a retail T3RS and he was experiencing exactly the same problems with three different CPUs – He also tried some Cellshock Blue, which use Micron D9 ICs and the results were the same.

Both Chris and I felt that a BIOS update will potentially fix the FSB limitation, but sadly the BIOS hasn’t arrived during the review.

Recovery from a bad overclock is another issue that needs addressing with the T3RS. When going for a high FSB, it’s inevitable that you are going to go too far.

When this happens, ideally, the board will reset itself, reboot, and possibly load the last known, safe settings – Unfortunately, this wasn’t the case and even the provided recovery measures wouldn’t work – In theory, you can press the home key to reset the FSB, or the insert key to reset all the BIOS settings. Nine times out of ten, I had to do a full CMOS clear before the board would boot, and when changing RAM types, I had to flip the battery out before it would post.

Also, be aware that the behaviour that I’m describing only happens at the edge of the board’s capabilities and as the BIOS matures, I have no doubt that this will improve.

The final problem and one we have heard a lot about on the forums is that the shipping BIOS will not allow the board to boot when there are CAS 7 modules installed. As far as we can see, there is no workaround for this, so be aware that if you have CAS 7 memory, you will have to have the BIOS pre flashed or use different sticks to flash to the later BIOS – once the BIOS is flashed, CAS 7 modules work fine.

The Good News.

Let’s be absolutely clear though, the X48-T3RS is no slouch, in fact, it clocks very well, and was extremely stable right up to 488x9.5 with the E8500!

488x9.5 @ 1.53v - 4561Mhz CPU - 1952Mhz RAM @ 1.82v - 9-9-9-24 - 1:2

The on board cooling is excellent and by far the best stock cooling I've seen on any board to date. I really like the way that a waterblock can be fitted to the North Bridge while still using the Thermalright heat pipe set up.

Update: BIOS 0725 manages to improve on some of the problems such as the overclock recovery and FSB limitation, but only by a whisker.

FSB wise, I was able to do this - Much better, but still not great for this kind of board.

The board layout is near perfect and has been tweaked from the original DDR2 design, so the previous gripes have now been fixed.

To sum up, the DFI UT X48-T3RS has had a hard act to follow with the X48-T2R, and in some respects such as cooling and layout, it builds on that reputation, but it does fall short in some key areas, namely sheer FSB and overclock recovery.

I will recommend the T3RS, but with some reservations – If you have a higher multi CPU, if you aren’t going for the absolute highest FSB & highest memory clocks possible, and you want a good solid board that will overclock well to relatively high levels, then the T3RS could be the one for you.

Pros