AM4 mobo's BIOS performance bias?!

[Liltrekkie]

1 hour ago, zMeul said:

-snip-

You look for any reason to shit on AMD don't you?

 

Do you even know what this setting does?

 

I'll give you a hint, someone in this thread told you what it does. It's borrowed from the enterprise market. 

 

Allow me to explain, Most OS's can control power usage and throttle the CPU down in light loads to conserve power. However in the server market, there is a BIOS setting that allows you to override those power and throttle settings in the OS. Typically you'll have FOUR (4) settings. Balanced, which detects CPU usage and throttles when  it can to reduce power but fully maintaining full power as necessary. Power saver, or some kind of reduced power state, will throttle the CPU and undervolt it as much as possible and only give the OS and applications the juice it needs. And then lastly, high performance, which is that, itll override power settings and maintain maximum processor performance, regardless of CPU usage. 

 

Go into windows power profiles then to advanced settings, change a plans settings, again hit advanced settings and look for processor power management. All 3 windows power plans have different processor power states. This AMD AM4 board is no different. They are exclusively press settings, well supposed to be in order to maintain maximum power and clock speeds so that an OS's power management doesn't skew the results. 

 

And I know youll find some excuse to blame AMD, you always will. But seriously, stuff it.

[done12many2]

2 minutes ago, SamStrecker said:

What that is was OCed??? What am I missing

 

Speculation over whether or not there is some type of performance enhancement option for use with specific benchmarks.  You know, boosting the result?

 

 

[zMeul]

18 minutes ago, Curufinwe_wins said:

Actually a well designed game would make extensive (almost extreme) use of AVX code... But most of them are not well designed...

you still need standard workloads for a lot of other subcomponents of the game

[Liltrekkie]

1 minute ago, done12many2 said:

-snip-

 

Except it just controls power management. Read my previous post explaining it, you'll understand. 

[MakeAMDGreatAgain]

8 minutes ago, Liltrekkie said:

Edit: Removed nvm wrong post srry

[Liltrekkie]

Just now, MakeAMDGreatAgain said:

I wonder how much Intel is paying you to shill on the LTT forums

 

I wish someone was paying me. I just call out BS as I see it. 

[done12many2]

1 minute ago, Liltrekkie said:

You look for any reason to shit on AMD don't you?

 

Do you even know what this setting does?

 

I'll give you a hint, someone in this thread told you what it does. It's borrowed from the enterprise market. 

 

Allow me to explain, Most OS's can control power usage and throttle the CPU down in light loads to conserve power. However in the server market, there is a BIOS setting that allows you to override those power and throttle settings in the OS. Typically you'll have FOUR (4) settings. Balanced, which detects CPU usage and throttles when  it can to reduce power but fully maintaining full power as necessary. Power saver, or some kind of reduced power state, will throttle the CPU and undervolt it as much as possible and only give the OS and applications the juice it needs. And then lastly, high performance, which is that, itll override power settings and maintain maximum processor performance, regardless of CPU usage. 

 

Go into windows power profiles then to advanced settings, change a plans settings, again hit advanced settings and look for processor power management. All 3 windows power plans have different processor power states. This AMD AM4 board is no different. They are exclusively press settings, well supposed to be in order to maintain maximum power and clock speeds so that an OS's power management doesn't skew the results. 

 

And I know youll find some excuse to blame AMD, you always will. But seriously, stuff it.

1 minute ago, Liltrekkie said:

 

Except it just controls power management. Read my previous post explaining it, you'll understand. 

 

 

You just explained Windows Power Options which is a very well-known feature.  What are the benchmarked label features for in the BIOS screen capture?  That's the topic at hand.  Not Windows Power savings features.

 

You just talked about elephants while everyone else here is talking about bears.

 

Why do you need benchmark specific power management?

[zMeul]

17 minutes ago, Ryujin2003 said:

Because many h8ters are going to say this is why Ryzen performed as it did during their benchmarks in order to boost sales, essentially lying to the consumer base, and Intel is still supreme. *add salt here*

don't you find it hypocritical? I do 

[Liltrekkie]

2 minutes ago, done12many2 said:

 

 

You just explained Windows Power Options which is a very well-known feature.  What are the benchmarked label features for in the BIOS screen capture?  That's the topic at hand.  Not Windows Power savings features.

 

You just talked about elephants while everyone else here is talking about bears.

 

Why do you need benchmark specific power management?

 

You skipped my entire explanation. I wasn't talking about windows power management though I did reference it for comparison . I was talking about a BIOS setting that would override Windows power management, keeping the processor from going into a low powerstate and throttling back to conserve power. 

 

This has been an enterprise setting for a long long time. 

[done12many2]

Just now, Liltrekkie said:

 

You skipped my entire explanation. I wasn't talking about windows power management though I did reference it for compassion. I was talking about a BIOS setting that would override Windows power management, keeping the processor from going into a low powerstate and throttling back to conserve power. 

 

This has been an enterprise setting for a long long time. 

 

I get that part.  Why would that feature, if it was indeed that, be specific and different for each benchmark labeled?

[Curufinwe_wins]

6 minutes ago, zMeul said:

you still need standard workloads for a lot of other subcomponents of the game

Yes, but that doesn't explain away the fact that games are coded like absolute dogshit compared to the potential offered by modern cpu's.

 

Also... most good hpc computing sees extreme use of AVX.

 

But anyways... the point remains.... this looks really sketchy as balls, and we will have to see exactly how big of a deal it is.

[Megah3rtz]

1 hour ago, nerdslayer1 said:

 in Canada most people don't speak french.   

I went to South Carolina for a vacation and somebody asked me how I spoke English so well.. 

[Liltrekkie]

Just now, done12many2 said:

 

I get that part.  Why would that feature, if it was indeed that, be specific and different for each benchmark labeled?

 

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.  

[zMeul]

Just now, Curufinwe_wins said:

Yes, but that doesn't explain away the fact that games are coded like absolute dogshit compared to the potential offered by modern cpu's.

because developers hire script kiddies and not programmers

because they cut development time to meet preposterous deadlines

.. pick your posion

[zMeul]

1 minute ago, Megah3rtz said:

I went to South Carolina for a vacation and somebody asked me how I spoke English so well.. 

because you're french 

[done12many2]

3 minutes 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.  

 

Sure, but what that tells me is that they have tuned the chip to its theoretical limit should a benchmark specific power savings feature be required to reduce voltage at the onset of a REAL load.  By the way, if that chip is so tuned to its limit that anyone of the benchmarks listed causes an overvolt/overcurrent type situation, Ryzen is not looking good at all.

 

I say this because none of those benchmarks listed produce anything close to what I consider a load.

[Doobeedoo]

Interesting. Also can't wait to see more info about mobos too.

[Liltrekkie]

1 minute ago, done12many2 said:

 

Sure, but what that tells me is that they have tuned the chip to it's theoretical limit should a benchmark specific power savings feature be required to reduce voltage at the onset of a REAL load.  By the way, if that chip is so tuned to it's limit than anyone of the benchmarks listed causes an overvolt/overcurrent type situation, Ryzen is not looking good at all.

 

It's not that they tuned the chip to it's limit. 

 

Look up prime95 haswell, there was something about the Prime95 AVX2 codec that caused it to send way too much voltage to the CPU, potentially causing damage. Server processors actually downclock when AVX2 is being used so no overvolting can occur. 

 

Like I said, there are very specific use cases when powerlimits need to be established at the BIOS level to maintain a safe system, thus averting any damage during a stress test or benchmark. 

[done12many2]

5 minutes ago, Liltrekkie said:

 

It's not that they tuned the chip to it's limit. 

 

Look up prime95 haswell, there was something about the Prime95 AVX2 codec that caused it to send way too much voltage to the CPU, potentially causing damage. Server processors actually downclock when AVX2 is being used so no overvolting can occur. 

 

Like I said, there are very specific use cases when powerlimits need to be established at the BIOS level to maintain a safe system, thus averting any damage during a stress test or benchmark. 

 

Once again, I already know this. 

 

AVX2 Pime95 loads and Cinebench R15 loads are NOTHING alike.  If you need a special power reducing feature to kick in when you run a Cinebench R15 load, you're going to need it for a lot of other loads throughout a days worth of computing.  R15 is not a heavy load by any stretch. 

 

I should add, that the issue with Haswell and Prime95 is not an issue of voltage, but current.  People mix this up all the time.  Voltage is just the potential, not the real issue.

[mattebad]

1 hour ago, SamStrecker said:

Yeah but why is this news?

Because it's Zmeul and he's upset AMD hasn't burned to the ground yet

 

But in all seriousness, this is on MSI and Asus, AMD has no part in mobo bios driver features, especially these tailored for the press with Auto overclock on. Bad MSI and Asus, stop being sketchy for no real reason

[mattebad]

1 hour ago, zMeul said:

it can detect workload

 

my mobo has a similar thing for 3DMark, disabled by default

Wait really? I was under the impression that was just C-states and not the mobo OC program. Mobo oc, for my asus board at least is strictly OC only, it doesn't scale the OC to the workload being done, I don't even think something like that exists, it' would just be C-states downclocking when workload is not intense.

[Liltrekkie]

1 minute ago, done12many2 said:

 

Once again, I already know this. 

 

AVX2 Pime95 loads and Cinebench R15 loads are NOTHING alike.  If you need a special power reducing feature to kick in when you run a Cinebench R15 load, you're going to need it for a lot of other loads throughout a days worth of computing.  R15 is not a heavy load by any stretch. 

Again, it may not be that its a power reducing feature, it may just be keeping the processor state at 100%, instead of letting the OS maintain processor state. 

 

Both choices are equally plausible. Or it could be some combination of both. 

 

We don't know how these Ryzen CPU's react to certain use cases. It's entire possible that they process certain workloads differently, thus need power reducing settings. Or it's just possible they want to over ride OS power management and keep everything at 100% for best results. 

 

Both are likely. 

[BuckGup]

25 minutes ago, done12many2 said:

 

Speculation over whether or not there is some type of performance enhancement option for use with specific benchmarks.  You know, boosting the result?

 

But it's still the CPU and consumers can acheive the same results. If they were using custom versions of the benchmark software or something more fishy then we would have a problem

 

[done12many2]

Just now, Liltrekkie said:

Again, it may not be that its a power reducing feature, it may just be keeping the processor state at 100%, instead of letting the OS maintain processor state. 

 

Both choices are equally plausible. Or it could be some combination of both. 

 

We don't know how these Ryzen CPU's react to certain use cases. It's entire possible that they process certain workloads differently, thus need power reducing settings. Or it's just possible they want to over ride OS power management and keep everything at 100% for best results. 

 

Both are likely. 

 

I get what you're saying.  I'm just not sure that we agree on why that feature would need to be specifically curtailed for each benchmark listed.  

 

It'll all be clarified soon enough so until then...  

[Liltrekkie]

3 minutes ago, done12many2 said:

 

I get what you're saying.  I'm just not sure that we agree on why that feature would need to be specifically curtailed for each benchmark listed.  

 

It'll all be clarified soon enough so until then...  

 

For a press CPU thing such as this it makes sense. It takes away any variables with scores. One run Windows might say "meh all you need is 80% clock speed" and another run it gives you 100% of your clock speed. 

 

Change your windows power settings and allow the minimum processor state to be 0%, then go to task manager and look at the performance tab.  You'll see the clock speed jump around depending on your task. Just writing this post my AMD x6 1055t is jumping between 1.2 GHZ to 3.2GHZ (it's max bclock OC speed) And itll continue to jump around even during a benchmark or gaming.