AM4 mobo's BIOS performance bias?!

[leadeater]

34 minutes ago, Misanthrope said:

What you're saying is that you definitively need to show evidence that the name means nothing

No what I was saying was show evidence it only applies to the named benchmark. Otherwise totally agree with what you said.

[Urishima]

13 minutes ago, Bouzoo said:

Interesting. Can't wait for TPU, Guru3D and Anandtech to strip them open. 

Word is that some reviewers are receiving BIOS updates at the last minute, forcing them to retest everything.

[matrix07012]

 

[zMeul]

39 minutes ago, leadeater said:

You're missing the point, show me it effecting the benchmark and only the benchmark and no other application.

I'm absolutely not missing the point

I'm providing as much info I can find and as it surfaces - I provided link to the BIOS that was originally thought it was "press only"; provided another screenshot from a 3rd party showing the setting being there; provided proof from two sources that AMD (!!), not ASUS, changed testing methodology just a day prior to release

the shit keeps piling up

 

 

where were you when the WCCFTech article was posted and couple of people, me included asked for actual evidence and not just allegations from Demerjian?

 

 

those double standards ........

[leadeater]

19 minutes ago, Dabombinable said:

Hint hint: chipset drivers.

Hmm that would be rather invasive drivers. I'd still like to see benchmarks of each setting and how it effects other applications before stating that it only applies to those specific benchmarks only.

[Dabombinable]

8 minutes ago, leadeater said:

Hmm that would be rather invasive drivers. I'd still like to see benchmarks of each setting and how it effects other applications before stating that it only applies to those specific benchmarks only.

You do realise how deeply embedded+important to systems functionality the chipset driver is?

[MageTank]

9 hours ago, Liltrekkie said:

 

Because there are some benchmarks that may cause a processor to send too much voltage in very specific use cases. See the Prime95 debate for example.  

Except that's not an issue with Prime95. That's an issue with AVX and LLC settings. You can get the same "excessive vcore" result with Linpack. That Prime95 debate was slain years ago, by people that actually know what they are talking about, lol. 

 

1 hour ago, LAwLz said:

-snip-

I like you.

 

So, while it pains me to defend the anti-AMD crusader, his intentions (aside from still wanting to see AMD burn) are noble here. Consumers need to know this information BEFORE benchmarks surface. Even if we do not know what these settings do, we need to know that there is the potential for deceit in these specific benchmarks.

 

Whether or not this is AMD's fault is to be seen, but I can say ASUS has done this many times before. ROG Memory Scramble allows people to use completely unstable memory overclocks to bench without crashing. Why are you not defending my memory scores against these people, @zMeul?! Don't let my ram get overshadowed by pseudo-stable overclockers. Crusade for justice!

[leadeater]

12 hours ago, zMeul said:

hardly the same thing since this settings are specifically tailored for select benchmarks

 

3 hours ago, leadeater said:

Different benchmarks favor different things, if something stops benefiting from more than 4-6 cores you could create a bios setting to preference boosting only that many cores there by increasing the benchmark scores. That very same setting could then also be used for non heavily threaded games.

 

I get that the names incite concern what I don't agree with is not being able to wait a very small amount of time before throwing around accusations.

 

2 hours ago, zMeul said:

these settings are tailored specifically for benchmarks, not for regular loads - they even bothered to name them as such

 

I get that you find it concerning about what those settings mean and the implications behind them, I find it very presumptuous that you can state with absolute certainty that those setting can and will only apply to the named benchmarks.

 

I object to the way you are presenting the information not the information itself. Without seeing the benchmarks, that include testing other applications with those options set, we can't know for sure that those setting only apply to the named benchmarks. If this is the case, however unlikely I think that is, then benchmarks with the setting disabled is the one we should use to compare to Intel systems to evaluate the platform differences and value of them.

 

When the benchmarks come out and we find that there are no results given with the setting disabled we will have to request one or all reviewers do that and add the results to the review. Where I will raise my alarm bells is if they refuse to in this case as that basically means they are not allowed to release those numbers.

[leadeater]

18 minutes ago, Dabombinable said:

You do realise how deeply embedded+important to systems functionality the chipset driver is?

Yes but chipset drivers don't monitor your running applications and then modify your bios/CPU setting when it detects certain applications. That would be some next level bullshit if Ryzen chipset drivers did that.

 

 

[TOMPPIX]

some guy on reddit got his 1700X "early"  someone posted that already oops

 

[TOMPPIX]

guys.... this might have something todo with this. 

http://www.amd.com/en-us/innovations/software-technologies/zen-cpu#neuralnet

[Fetzie]

1 hour ago, leadeater said:

Th.

 

Bios's have been isolated from OS's for a very long time for virus and malware protection. If you think this has changed I'd be more worried about system compromises than skewed benchmarks

 

UEFI bioses are allowed to expose components to software running on the OS via APIs.

[Dabombinable]

27 minutes ago, leadeater said:

Yes but chipset drivers don't monitor your running applications and then modify your bios/CPU setting when it detects certain applications. That would be some next level bullshit if Ryzen chipset drivers did that.

 

 

You mean current chipset drivers don't. Remember, this entire platform from AMD shares very little with all of their past ones, and all the drivers would have to do is tell the BIOS which of the benchmarks is run, and thus trigger the preset which gives the best performance in it (with manually setting it negating the need for the chipset driver to communicate).

[Dabombinable]

52 minutes ago, MageTank said:

Except that's not an issue with Prime95. That's an issue with AVX and LLC settings. You can get the same "excessive vcore" result with Linpack. That Prime95 debate was slain years ago, by people that actually know what they are talking about, lol. 

 

 

I like you.

 

So, while it pains me to defend the anti-AMD crusader, his intentions (aside from still wanting to see AMD burn) are noble here. Consumers need to know this information BEFORE benchmarks surface. Even if we do not know what these settings do, we need to know that there is the potential for deceit in these specific benchmarks.

 

Whether or not this is AMD's fault is to be seen, but I can say ASUS has done this many times before. ROG Memory Scramble allows people to use completely unstable memory overclocks to bench without crashing. Why are you not defending my memory scores against these people, @zMeul?! Don't let my ram get overshadowed by pseudo-stable overclockers. Crusade for justice!

I could use that memory scramble on my P5Q Deluxe-its proven impossible to get the IMC to run stable with 2x 2GB sticks of DDR2 1066 (it needs 1x2GB and 2x1GB in asymetric dual channel-which increases the latency by 10-20ns, else its limited to just over 900MHz and 800MHz flat with all 4 slots filled)

[leadeater]

9 minutes ago, Dabombinable said:

You mean current chipset drivers don't. Remember, this entire platform from AMD shares very little with all of their past ones, and all the drivers would have to do is tell the BIOS which of the benchmarks is run, and thus trigger the preset which gives the best performance in it (with manually setting it negating the need for the chipset driver to communicate).

Would certainly make me consider not using Ryzen if drivers were doing application monitoring, at least with an annoying Asus Ai Suite application I can uninstall/not install it.

 

Edit:

Not that I actually plan on using Ryzen, got too many PCIe cards for the number of lanes.

[Dabombinable]

13 minutes ago, leadeater said:

Would certainly make me consider not using Ryzen if drivers were doing application monitoring, at least with an annoying Asus Ai Suite application I can uninstall/not install it.

 

Edit:

Not that I actually plan on using Ryzen, got too many PCIe cards for the number of lanes.

I was toying with the idea mysel. But with my GTX 970 already being damaged and running at PCIe x8, I might as well just stick with my 4790K and get a second GTX 970 (or a new single card with at least 6GB vRAM)

[leadeater]

Just now, Dabombinable said:

I was toying with the idea mysel. But with my GTX 970 already being damaged and running at PCIe x8, I might as well just stick with my 4790K and get a second GTX 970 (or a new single card with at least 6GB vRAM)

Yea I think for the majority of people getting a new GPU is a much wiser spend than buying a new CPU + MB + RAM (if DDR3), no matter how good Ryzen actually is.

 

Also I had one of my 290X's doing the same x8 thing, re-seated it and fiddled with a few CPU and ram OC settings and voltages got it back to x16. Happened right on updating to 17.2.1 drivers so not sure what to blame.

[done12many2]

17 minutes ago, leadeater said:

Would certainly make me consider not using Ryzen if drivers were doing application monitoring, at least with an annoying Asus Ai Suite application I can uninstall/not install it.

 

Edit:

Not that I actually plan on using Ryzen, got too many PCIe cards for the number of lanes.

 

I wish I could install the fan controller alone from the AI Suite.  They can keep the rest of that garbage.  

 

The monitoring portion of the AI Suite is so resource dependent that it's useless when it matters the most.  When a CPU is loaded.

[leadeater]

1 minute ago, done12many2 said:

 

I wish I could install the fan controller alone from the AI Suite.  They can keep the rest of that garbage.  

 

The monitoring portion of the AI Suite is so resource dependent that it's useless when it matters the most.  When a CPU is loaded.

The R4BE has the OC panel I can mount in a 5.25 bay. Can adjust fans and OC presets from that no applications required , a lot of Asus boards have the plug for it but weren't sold with it or listed it as an option but you can just get one off ebay maybe.

[Dabombinable]

7 minutes ago, leadeater said:

Yea I think for the majority of people getting a new GPU is a much wiser spend than buying a new CPU + MB + RAM (if DDR3), no matter how good Ryzen actually is.

 

Also I had one of my 290X's doing the same x8 thing, re-seated it and fiddled with a few CPU and ram OC settings and voltages got it back to x16. Happened right on updating to 17.2.1 drivers so not sure what to blame.

No fiddling will fix my 970-it runs at x8 in any motherboard. And it at one point had the smell of fried electronics coming off it (with a visual inspection showing nothing damaged).

1 minute ago, leadeater said:

The R4BE has the OC panel I can mount in a 5.25 bay. Can adjust fans and OC presets from that no applications required , a lot of Asus boards have the plug for it but weren't sold with it or listed it as an option but you can just get one off ebay maybe.

I wonder if my Sabertooth has the connectors required.

[Dabombinable]

@leadeater

The connectors are "TPM" and "AAFP"?

[19_blackie_73]

as long as these settings are available for everyone, I wouldn't bother.

[leadeater]

4 minutes ago, Dabombinable said:

@leadeater

The connectors are "TPM" and "AAFP"?

ROG_EXT

 

AAFP is the front panel audio,

[done12many2]

3 minutes ago, leadeater said:

The R4BE has the OC panel I can mount in a 5.25 bay. Can adjust fans and OC presets from that no applications required , a lot of Asus boards have the plug for it but weren't sold with it or listed it as an option but you can just get one off ebay maybe.

 

I know exactly the "plug" you're talking about.   

 

I was just saying the AI Suite sucks, but I like the fan controller portion because by default and left untouched, the presets trigger firmware level fan settings preset in BIOS/UEFI. Fan control doesn't go to shit when controlled by firmware like it does with AI Suite overall once you place a decent load on your system.

 

 

 

 

[leadeater]

1 minute ago, done12many2 said:

I know exactly the "plug" you're talking about.   

Your OC panel is so much nicer than my old shitty one . Wanna swap? lol