Asus TUF X299 Mark 2 BIOS

[batboy]

A new BIOS update is available for the TUF X299 Mark 2 motherboard as well as for other models too. Anyone try version 1004 yet? I'm gunshy because I updated to version 0802 as soon as it was released and then I found out it sucked.


EDIT: Check out post #15 for good news about memory overclocking with the new BIOS version 1004.

 

[EarthDog]

0802 worked fine for me in the Mark 2 when I reviewed it a couple weeks back. I haven't tried 1004.

What was wrong with it?

 

[batboy]

My biggest gripe was that it claimed to improve M-2 NVMe drive performance, but they lied. I had some significant decreases in performance. I posted before and after screenshots of Crystal Disk benchmark in another thread back when 0802 first came out.

I saw there was a new BIOS out for my Asus TUF X299 Mark 2 (version 0802 released 9/15/2017). Here are the changes:

TUF X299 MARK 2 BIOS 0802
1.Fixed SanDisk M.2 device issue.
2.Fixed PLEXTOR device issue.
3.Update Intel X-series CPU (6-core and above) microcode.
4.Improved DRAM / system compatibility.
5.Improved secure erase function.
6.Improved Q-Fan function.
7.Improved system performance.
8.Improve PCH PCIe SSD Performance

Cool, look at #8 "Improve PCH PCIe SSD Performance." I download the BIOS update file and flash it. I'm rubbing my hands with glee, my M.2 RAID 0 will be even faster.

Here is the first bench (old BIOS):

[

Results after flashing to ver. 0802:



Where's my improved performance? Overall, I think scores are down except for a couple. Give me back my freakin' performance.

 

[Woomack]

I think I mentioned that 0802 is worse than previous version just after it was released. The same stability etc but somehow a bit worse OC. I had no issues with storage performance on my drives but nothing was better. I guess I won't check new BIOS as my TUF went on auction and is already packed.

 

[mackerel]

I hope they included the ME bugfix as I thought they were rather late on that given this is a current mobo. It isn't explicitly mentioned in the change notes. I'll try it out when I can but might not be today.

 

[batboy]

ME bugfix? I had to google what ME was, looks like it is the Intel Management Engine. First I heard of this bug.

I decided to bite the bullet and I updated the BIOS to version 1004. So far, I'm having trouble getting the RAM to run at DDR4-4000 like I had before. In fact, I haven't even got it to run at DDR4-3600 yet and that's what it's rated at. Anyway, I looked in the device manager and saw I have Intel Management Engine driver version 11.7..0.1014 (which appears to be the most recent update available). Someone else with this motherboard with the 0802 BIOS version take a look and see what ME version you have.

 

[EarthDog]

ME and BIOS are two separate things, but can be updated together via bios. You can run 0802 bios and have the latest ME driver, for example.

As far as nvme performance, it makes sense you didnt see any improvement as you are using raid and saturating the dmi bus in the first place. Im not surprised at your results. The write drop could have been anything...or because of the bios.

 

[batboy]

It was the BIOS, I ran Crystal, then flashed the BIOS, and then ran Crystal again. Nothing else done in between. Maybe it was because of the RAID 0? Dunno, but they stole my performance and I want it back.

I'm up to DDR4-3000 now. I'm gradually bumping up the RAM frequency... I enabled DRAM learning, so now I'm sort of trying to sneak the memory speed up a little at a time.

 

[mackerel]

https://www.intel.com/content/www/us/en/support/articles/000025619/software.html

Above link to tool to check if your system is vulnerable to recently reported ME bugs. Whatever was in bios 0802 was affected. It has to be patched at firmware level, not driver level, since ME operates below OS level. On my Asus Z170 boards, they didn't release a full bios, but just a ME update tool. In part that made me curious why they were taking so long on what is a current mobo.

 

[batboy]

Ok, I downloaded and ran that tool.

Based on the analysis performed by this tool: This system is vulnerable.

 

[batboy]

My hopes were dashed, disk performance is way worse.



Maybe it is because of the RAID? I'm tempted to go back to my single 960 Pro and see if that makes a difference.

 

[mackerel]

So they haven't included a fix with it... what are Asus doing? It's not that they can't, as they've implemented it on Z170 boards already.

 

[EarthDog]

My hopes were dashed, disk performance is way worse.

View attachment 195366

Maybe it is because of the RAID? I'm tempted to go back to my single 960 Pro and see if that makes a difference.

yes, you wont see much, if any at all improvement with R0 because it is saturating the dmi bus since at least one of those, if not both, are fed from the chipset.

Also, your screenshots show a drive double in space used from one run to the next. The latest screenshot shows it 3x more full, though it is only 25%. It really could be anything. Id bet good money says the bios didnt crush performance that much...

Also, some FW for the 961 crippled its performance. Did you flash it between those screenshots amd today???
EDIT: That was 960 pro...my fault.

If you really wanted to investigate the issue, this should be done on an empty drive which is Secure Erased between each run. The more you bench it, the slower they tend to get.

 

[batboy]

At least I'm able to run the RAM in quad channel at DDR4-3600 again. Like I said, when I tried to go straight to 3600, it would not even try to boot. This was the first time I had to push the "MemOK" button on the motherboard to clear the RAM settings. Not as extreme as clearing the CMOS. What I had to do was go up in smaller steps (2800, 3200. 3600). Now that I'm at DDR4-3600, it's the same exact settings that I tried the first time. Weird.

I know with SSDs as you increase the amount of files being stored, you gradually lose performance. But, no way it can be that much. I move a few directories over to my storage drive to free up space (at 15% now and it was 24%). I'll close everything and run Crystal Disk again. Both drives are at 100% health with no drive warnings and no drive failures according to HWinfo64.

 

[batboy]

Ok, I've whined and complained about the decrease in M-2 drive performance. But, I did discover one good thing about the BIOS update (version 1004). This is the first time I could do DDR4-4200 at CL16 in quad channel. I got a screen shot once with the 7740X running DDR4-4200 with CL17 in dual channel (but it was not completely stable).

 

[Woomack]

Is it working in benchmarks or you made only clock screenshot ?

 

[batboy]

Phooey, the settings in the screenshot were not stable when I tried to benchmark. I changed the timings to 17-17-17-36 and I bumped up DIMM voltage, which seems to be working better. I'll post back with results later tonight.

EDIT: Ram at DDR4-4200 17-17-17-36 quad channel is stable in a few of the benches I ran so far. In fact, I just got a new personal best in Geekbench 3 single core and I wasn't even trying. Going to try bumping stuff up and really giving it a go.

EDIT: Well, 4200 is not completely stable even when I tried CL18 and even more voltage. I can run it when the CPU is at 4.6 gig, but if I try to bench at 4.8 gig, I blue screen (which is usually RAM). I can bench at 4.8 and DDR4-4000 (16-16-16-36) all day long. So, seems a bit strange that I can't get 4200 completely stable. That said, I still think the BIOS helped memory overclocking, because before the BIOS update, there was no way it would even boot at 4200 (quad channel), so we're making progress.

 

[Woomack]

I had the same at 4100-4200 on TUF and finally stabilized it at 18-20-20 1.35V. On Taichi XE there is only 1 official BIOS so I guess it's still time for improvements. Taichi XE OC better using 16GB modules and stabilizes easier at 4000 using 8GB modules but can't pass 4100 in benchmarks while TUF could. On Taichi, 64GB kit runs at 3733 fully stable and boots at 4000 for screenshots while on TUF there were random issues to boot at 3733+ not to mention anything higher.
Yesterday I've moved to X299 ITX but I'm still waiting on 4x8GB kit. I wasn't thinking about this before but if I'm right then all 8GB and higher capacity SODIMM modules are dual rank.

 

[EarthDog]

On Taichi XE there is only 1 official BIOS so I guess it's still time for improvements.

Considering the only difference between the XE and non-XE Taichi is the heatsink, it has all the other Taichi BIOS rollups as well. At least it should. These two boards should not behave different at all.

 

[Woomack]

I found 2nd difference, it's the 2nd 8 pin power connector ... I told you there is something more I'm not sure if they used all fixes from non-XE Taichi, I don't know how it's overclocking and it's hard to compare them when I had no chance to test both with the same memory. I don't expect big difference though. I've received beta BIOS from support but it works the same as official one so maybe it's max for these boards, hard to say. For sure I'm not complaining. Without any additional training options it works above 4000 in quad channel.