Playing with the AGESA-1004a BIOS

D

DuronBurgerMan

[H]ard|Gawd
Joined
Mar 13, 2017
Messages
1,340
So for the Asus X370 Prime, the new AGESA-1004a BIOS is out. I thought I'd give it a try.

So far, the results are mixed:

My memory, which was running at 2933, now only clocks up to 2400. 2666 and 2933 will not POST, even with higher SOC and RAM voltages. This is a bummer.

Cinebench scores are up a bit, so even running the RAM slower, the system is faster? Or maybe AMD put some other things in this update. Weird.

On the positive side, boot time has been reduced another 5 seconds. Now at 19 seconds total.

I may roll back to the 515 BIOS, though. The RAM compatibility & speed should improve, not get worse, lol. Maybe AMD pushed this update too quick to get in under the wire for the Ryzen 5 launch.

How has the update worked out for everyone else?
 
CH6 Hero - 0802 here.

My memory still only works at 2666, but everything is absolutely faster. There's supposed to be up to 5% performance increase due to microcode, but my 960 Evo benchmarks at roughly 10% faster.
 
I don't think this AGESA update had much to do with the IMC, but more towards CPU related issues.

Quote from a PM the Gigabyte rep sent me
The new AGESA code didn't actually do anything for the memory controller though, AMD rolled this out to deal with other CPU issues
Click to expand...

With that said, my Gigabyte K7 is running fine, Ram is running at 3200Mhz on the latest BIOS.
 
THUMPer said:
I don't think this AGESA update had much to do with the IMC, but more towards CPU related issues.

Quote from a PM the Gigabyte rep sent me

With that said, my Gigabyte K7 is running fine, Ram is running at 3200Mhz on the latest BIOS.
Click to expand...

It's odd, then, that it would have an effect on my memory speeds.
 
dook43 said:
CH6 Hero - 0802 here.

My memory still only works at 2666, but everything is absolutely faster. There's supposed to be up to 5% performance increase due to microcode, but my 960 Evo benchmarks at roughly 10% faster.
Click to expand...

That would explain why my RAM is now slower, but everything benches faster anyway.
 
Hey, as another positive note, Ryzen Master now reports the right temps. No +20C offset.

Now if only AMD and/or Asus could re-enable the fast memory clock rate I had before, it'd be a winner.

As it stands, I'll tolerate it, for now, and won't roll back to 515.
 
DuronBurgerMan said:
So for the Asus X370 Prime, the new AGESA-1004a BIOS is out. I thought I'd give it a try.

So far, the results are mixed:

My memory, which was running at 2933, now only clocks up to 2400. 2666 and 2933 will not POST, even with higher SOC and RAM voltages. This is a bummer.

Cinebench scores are up a bit, so even running the RAM slower, the system is faster? Or maybe AMD put some other things in this update. Weird.

On the positive side, boot time has been reduced another 5 seconds. Now at 19 seconds total.

I may roll back to the 515 BIOS, though. The RAM compatibility & speed should improve, not get worse, lol. Maybe AMD pushed this update too quick to get in under the wire for the Ryzen 5 launch.

How has the update worked out for everyone else?
Click to expand...

Can't roll back the AGESA update. It also ruined RAM OC for me. Can't run faster than 2133.
 
LigTasm said:
Can't roll back the AGESA update. It also ruined RAM OC for me. Can't run faster than 2133.
Click to expand...
This update was not rolled into a BIOS, as in, it was some sort of stand-alone microcode update that you applied?

If not, then flashing to a previous BIOS version should indeed "roll back" the AGESA, since they are self-contained modules inside the BIOS package. That's why AMI's MMTool (for example) has a Microcode option, but so do other tools. People have ripped out updated Intel microcode in one board vendor's BIOS to package into their own, when their vendor hasn't offered an update (or dropped support entirely.)

Nevertheless, I think 1.0.0.4B is the one AMD was referring to when they spoke of good things to come. Seems like vendors were just appeasing the masses by tossing out ones with 1.0.0.4A, since for any non-enthusiast or anyone with modules already stuck at 2400 speeds, it would offer performance gains.
 
My RAM still works at 3200 CL16. Still having cold boot problems at 3200 CL14 due to the lack of a DRAM Boot Voltage option on the Prime (initial DRAM training always happens at 1.2V, which is just one big no-no at 3200 CL14 and faster, even with B-die).

For those having problems, there's unfortunately no way to roll back. It will just refuse to flash and say the BIOS is invalid.
 
JimmiG said:
My RAM still works at 3200 CL16. Still having cold boot problems at 3200 CL14 due to the lack of a DRAM Boot Voltage option on the Prime (initial DRAM training always happens at 1.2V, which is just one big no-no at 3200 CL14 and faster, even with B-die).

For those having problems, there's unfortunately no way to roll back. It will just refuse to flash and say the BIOS is invalid.
Click to expand...

I'm mildly annoyed. But overall system performance improved 2-3% in all tested benchmarks despite the slower RAM clock, so I guess I really shouldn't complain. Whatever else came with this update seemed to give a nice overall boost to the entire system.
 
Formula.350 said:
This update was not rolled into a BIOS, as in, it was some sort of stand-alone microcode update that you applied?
Click to expand...

Yeah, it was rolled in to a BIOS update. I did try rolling back the BIOS but the effects were the same once the microcode update was applied. This was on a B350 Prime board so I don't know if it was rolled out the same way on all boards
 
LigTasm said:
Yeah, it was rolled in to a BIOS update. I did try rolling back the BIOS but the effects were the same once the microcode update was applied. This was on a B350 Prime board so I don't know if it was rolled out the same way on all boards
Click to expand...
You may have to do a complete flash where it overwrites all of the blocks. I wouldn't be surprise if that option isn't offered in the BIOS flash option. The "Emergency Recovery" method should (if the B350 models have it), otherwise using AMIAFU for DOS has the capability. However, the DOS method is more trouble than it's worth IMO. I'd be just gritting my teeth until another BIOS revision is kicked out. :\
 
Formula.350 said:
You may have to do a complete flash where it overwrites all of the blocks. I wouldn't be surprise if that option isn't offered in the BIOS flash option. The "Emergency Recovery" method should (if the B350 models have it), otherwise using AMIAFU for DOS has the capability. However, the DOS method is more trouble than it's worth IMO. I'd be just gritting my teeth until another BIOS revision is kicked out. :\
Click to expand...

Well, luckily my Crosshair VI showed up, the board I wanted in the first place. We'll see what that one does.
 
OC is a little more picky on Agesa 1004 too.

I had to up the voltage a tick, and use LLC to get it to pass the Intel Burn Test under this microcode and BIOS revision @ 3.95 GHz.
 
I finally got a few minutes to take my crap out the boxes and assemble it for testing. I'm running a Gigabyte AX370-Gaming K7 with a 1700x and a 16GB Corsair LPX 3200 C16 kit. I could not get the XMP profile to take even with a bump in VDIMM to 1.45 volts. I checked and the F3 bios based on AGESA-1004a is up. I flashed to it and got some joy when it posted with the XMP profile but it didn't hold on the reboot. It would randomly post at stock VDIMM and XMP timings but it was definitely not stable. After some tweaking I got consistent POSTs at 1.45v VDIMM so I bumped it to 1.47v to be sure and now I'm Memtest86 testing 40% with no errors so far!
 
Zareek said:
I finally got a few minutes to take my crap out the boxes and assemble it for testing. I'm running a Gigabyte AX370-Gaming K7 with a 1700x and a 16GB Corsair LPX 3200 C16 kit. I could not get the XMP profile to take even with a bump in VDIMM to 1.45 volts. I checked and the F3 bios based on AGESA-1004a is up. I flashed to it and got some joy when it posted with the XMP profile but it didn't hold on the reboot. It would randomly post at stock VDIMM and XMP timings but it was definitely not stable. After some tweaking I got consistent POSTs at 1.45v VDIMM so I bumped it to 1.47v to be sure and now I'm Memtest86 testing 40% with no errors so far!
Click to expand...

Thanks, I've been trying to get nearly the exact same spec'd rig to work for a week now with no joy.
 
DuronBurgerMan said:
So for the Asus X370 Prime, the new AGESA-1004a BIOS is out. I thought I'd give it a try.

So far, the results are mixed:

My memory, which was running at 2933, now only clocks up to 2400. 2666 and 2933 will not POST, even with higher SOC and RAM voltages. This is a bummer.

Cinebench scores are up a bit, so even running the RAM slower, the system is faster? Or maybe AMD put some other things in this update. Weird.

On the positive side, boot time has been reduced another 5 seconds. Now at 19 seconds total.

I may roll back to the 515 BIOS, though. The RAM compatibility & speed should improve, not get worse, lol. Maybe AMD pushed this update too quick to get in under the wire for the Ryzen 5 launch.

How has the update worked out for everyone else?
Click to expand...

Asus Prime B350M-A, got memory up from 2400 to 2667, so improvement for me.
 
JimmiG said:
My RAM still works at 3200 CL16. Still having cold boot problems at 3200 CL14 due to the lack of a DRAM Boot Voltage option on the Prime (initial DRAM training always happens at 1.2V, which is just one big no-no at 3200 CL14 and faster, even with B-die).

For those having problems, there's unfortunately no way to roll back. It will just refuse to flash and say the BIOS is invalid.
Click to expand...

If that is the case you will also have problems updating the bios. So what are you going to do, return the motherboard??? What is causing the cold boot problem?????
 
os2wiz said:
If that is the case you will also have problems updating the bios. So what are you going to do, return the motherboard??? What is causing the cold boot problem?????
Click to expand...

The bios is , if it is the same as on crosshair hero VI it could be that it does not allow enough retries for the ram to be validated (or still stuck at to low dram boot voltage).
 
You must log in or register to reply here.
Back
Top