AMD Announces BIOS Fix for Ryzen 3000 Boost Clocks - hits September 10th

Status
Not open for further replies.
Gideon said:
As soon as you use 2 cores on a Ryzen with any kind of load your going to slow to 4.4 or less no matter what so it's damn pointless to worry about
Click to expand...
As Algrim and I both state earlier in this thread, using Audacity and/or ffmpeg for transcoding, single-thread performance is important.
Perhaps we bought Ryzen systems for actual work and not just games, and need the additional threads for the cost, but would also like the advertised single-core boost speeds.

If you got a app that is totally dependent on 1 thread then why the hell did you buy a AMD chip in the first place, Intel clocks higher.
Click to expand...
Oh, you mean the Intel CPUs with the 20+ hardware exploits that may or may not have been patched with the, depending on the workload and task, massive 10-70% performance hits?
Sure, but clocks don't mean much at that point when the next exploit and patch are around the corner to destroy any semblance of performance originally paid for - SWAPGS just hit last month, and I wonder what will be found this month for Intel CPUs.

That's why my 6700K @ 4GHz now acts and performs like a 2600K @ 3.4GHz, if I get lucky with the workload... fool me once, and all that.
AMD is a company, and while they are kicking ass and taking names right now, they should still follow through and fix what isn't up to spec, and judging by the OP, they hopefully will.

you would know that if you used the thing
Click to expand...
We do use them, and know what the fuck we are talking about - at this point, I sure can't say the same for you.
Not everyone uses their CPUs exclusively for gaming, just FYI.
 
Red Falcon said:
Outside of fucking video games, single-thread performance is extremely important and still very relevant, regardless of the OS.
I'm amazed so many here think these CPUs are used just for video games and that the single-core boost clocks don't matter for anything else.

We aren't just talking about video games, we are talking about all applications, and video games are just a fraction of that.
The extreme ignorance in these threads is just astounding, hot damn.
Click to expand...
So tell me about all these applications that we are overlooking, please.
 
Red Falcon said:
We do use them, and know what the fuck we are talking about - at this point, I sure can't say the same for you.
Not everyone uses their CPUs exclusively for gaming, just FYI.
Click to expand...

Man been a really long time since 200 MHz really meant anything to me in a workload I was doing. I might need a Atomic clock to be able to measure the difference in time it took to complete. Unless you like synthetic benchmarks then yeah you can see a tiny difference in them. This whole boost thing is really overblown except for a tiny few that the chip is just not boosting right at all.
 
Gideon said:
Man been a really long time since 200 MHz really meant anything to me in a workload I was doing. I might need a Atomic clock to be able to measure the difference in time it took to complete. Unless you like synthetic benchmarks then yeah you can see a tiny difference in them. This whole boost thing is really overblown except for a tiny few that the chip is just not boosting right at all.
Click to expand...

Translation: No one should ever overclock because no one knows the difference anyway. I guess we should just ignore the URL of the site we're on... :whistle:
 
Red Falcon said:
There are still a shit ton of programs out there that are strictly single-threaded, or are lightly-threaded where having that single boost core would greatly benefit them.
Audacity transcoding, Windows updates, clamd, database applications, many updater programs (both Windows and Linux), certain compression tools, etc.

Believe it or not, your OS still has many back-end programs that are strictly single-threaded. ;)
Click to expand...
Lol, this is funny... Many background programs.... And Windows update? This is proof of needing 100 more mhz? Seriously, just say WinZip and so e outdated software can still use 1 thread. I understand there are some examples, and if these are the apps you use the most, you probably bought Intel anyways for better single thread performance. Outside of a couple of updaters and an unzip app, you won't really notice much. Possibly some older games? Most newer games use ate least a few threads even if it can't use 6+, 4 is pretty common.
This said, they probably should have said it would hit lower boost clocks and then let people be surprised to get more. Hopefully this fix gets people to the boost clocks, we can get some benches that show imperceivable gains and everyone can move on with their lives.
 
Dan_D said:
Actually, historically, when previous boost clocks were raised for some CPU's, the benchmark scores either didn't improve or went down. The frequencies going up isn't necessarily a guarantee that the actual performance will improve. If we were talking about a few hundred MHz, then naturally we would see consistent improvements in performance. However, when we are talking about clock speeds under 100MHz, I wouldn't bet on it. What really matters is how long they hold the boost clocks. That too is often a problem so unless the new BIOS addresses this, I wouldn't bet on any benchmark scores improving substantially.
Click to expand...
Yeah, they are going to boost an extra 100mhz for 10ms, then drop back 300mhz to make up for the thermals. Then we'll see scores drop across the board and everyone will be asking for the lower more consistent boosts :). This was all made up, but you could understand the logic I hope.
 
Let's see what the new BIOS update brings. If it doesn't fix the issue then AMD has a larger problem on their hands than anticipated.
 
Notable performance or not, its the precedence of this whole thing that's concerning. This is the same forum that went after Nvidia with razor sharp claws in recent memory but is now giving AMD a free pass for their own consumer unfriendly shenannigans.. 100mhz now sets the precedence for 500mhz next time (in fact, i must repeat that it IS ~400mhz or more for some 3900x users, counter arguments are lumping all users into the best case scenario to minimize the issue)

These chips are all binned, that's why we have a 3600x and 3800x. Look at silicon lotto results, virtually no 3700x will hit 3800x clocks. It should have been known that only one in 20 chips would see these boosts. They could have shipped a lower SKU or just printed reality on the box. "Up to" is nonsense, we're not advertising car insurance savings, put the clocks that the lowest denominator will reach on the box or split the sku.

I'm glad AMD is acknowledging it and trying to give us some boost back and I know they tried to push the chips to the absolute edge to get those competitive numbers, but in the end it's a bug or they just decided to be a bit unethical.
 
Algrim said:
Translation: No one should ever overclock because no one knows the difference anyway. I guess we should just ignore the URL of the site we're on... :whistle:
Click to expand...

And then a person, such as myself, responds with: R5 1600 at 4.0 GHz, R7 1700 at 3.8 GHz and R5 2600 at 4.0 GHz, all core and wonder where the person said anything about overclocking in the post you quoted. :whistle:
 
AceGoober said:
Let's see what the new BIOS update brings. If it doesn't fix the issue then AMD has a larger problem on their hands than anticipated.
Click to expand...

I rather a bios be released that improves performance without the need to clock higher than to clock higher but reduce or even have the same performance. But hey, folks are fretting about things and yet, we are supposed to be manual overclockers here. :)
 
ManofGod said:
I rather a bios be released that improves performance without the need to clock higher than to clock higher but reduce or even have the same performance. But hey, folks are fretting about things and yet, we are supposed to be manual overclockers here. :)
Click to expand...
Give me immaterial frequency numbers instead of performance anyday.
 
Red Falcon said:
As Algrim and I both state earlier in this thread, using Audacity and/or ffmpeg for transcoding, single-thread performance is important.
Perhaps we bought Ryzen systems for actual work and not just games, and need the additional threads for the cost, but would also like the advertised single-core boost speeds.
Click to expand...

So you're justifying at most a few second margin on 2 to 3 hundred dollar processors for unoptimized free software. Shenanigans.

Audacity is great software plagued by the inability to evolve. Just a few years ago they were still holding on to pre sse compatibility (aka pre pentium 3) They have finally moved past Visual Studio 2008 to I believe 2015, so we'll see. It's a niche area that should not be used to bolster such an obscure argument.
 
  • Like
Reactions: N4CR
like this
Everyone is complaining that they don't hit the MAX boost clock but are missing the point that 1- When running full load, all cores are overclocked and this gives the best user experience and 2- when not pegged, CPU is resting hence no need to boost.
But I understand why people feel cheated (although the spec is specifically named in a way to be considered a characteristic and not hard spec).

Please STOP looking at single core boost when all other cores are already overclocked... and yeah AMD, please find a better way to market this trivial Max boost thing you're printing everywhere.
I doubt this would get any legal traction but it's nice that AMD is owning it and promised to work on it, let's see what happens now.

We currently have no data on how they determined the STAR core and how this MAX boost is also determined. Do they test core one by one at factory giving an unfair thermal advantage ?
 
MMitch said:
Everyone is complaining that they don't hit the MAX boost clock but are missing the point that 1- When running full load, all cores are overclocked and this gives the best user experience and 2- when not pegged, CPU is resting hence no need to boost.
But I understand why people feel cheated (although the spec is specifically named in a way to be considered a characteristic and not hard spec).

Please STOP looking at single core boost when all other cores are already overclocked... and yeah AMD, please find a better way to market this trivial Max boost thing you're printing everywhere.
I doubt this would get any legal traction but it's nice that AMD is owning it and promised to work on it, let's see what happens now.

We currently have no data on how they determined the STAR core and how this MAX boost is also determined. Do they test core one by one at factory giving an unfair thermal advantage ?
Click to expand...

They must be using fairy magic from Disney or something to set those boosts:

83833F36-49C5-43DA-9099-58E5D5C84FE8.png


Given that’s all core, it’s a bit off mark, even going very sub zero in temps.

What I expect is if I custom cool a CPU I at least get an all core boost to the “max boost” rating, potentially hitting above. That has been the past precedent.

I am still skeptical this bios update will fix this given manual OC results but I suppose time will tell.

I’d love to see the same chart for single thread boosts if anyone can find it.
 
ManofGod said:
And then a person, such as myself, responds with: R5 1600 at 4.0 GHz, R7 1700 at 3.8 GHz and R5 2600 at 4.0 GHz, all core and wonder where the person said anything about overclocking in the post you quoted. :whistle:
Click to expand...

The person I quoted said 200 MHz wasn’t noticeable. With that attitude why overclock? You’re not going to notice it anyway.
 
Algrim said:
The person I quoted said 200 MHz wasn’t noticeable. With that attitude why overclock? You’re not going to notice it anyway.
Click to expand...

Single thread doesn’t matter either. Lets all start running cheap xeons and bulldozer.

It’s obviously true since the 3900x does better than the 9900k/KF in games.... /s
 
Dayaks said:
Single thread doesn’t matter either. Lets all start running cheap xeons and bulldozer.

It’s obviously true since the 3900x does better than the 9900k/KF in games.... /s
Click to expand...

At least boost clocks work better on Ryzen then your DLSS setting or the missing DLSS 2x on your Nvidia card, I remember you telling me that was the big reason you got the card. Yet here you are, shouldn't you be in a Nvidia forum raising hell about it? 200 MHz is unnoticeable except in a benchmark, but yeah that suddenly means I have 0 interest in overclocking. Back when I had 1 GHz processors or less then yeah that actually meant something. Take a 9900K at 4.8 then to 5.0 and tell me you actually notice a difference in how it runs. This whole thing is silly with false outrage, it's a boost clock and last time I saw that was not a guarantee you will hit them.
 
Dayaks said:
Single thread doesn’t matter either. Lets all start running cheap xeons and bulldozer.

It’s obviously true since the 3900x does better than the 9900k/KF in games.... /s
Click to expand...

For pure gaming I would compare against a 3700X anyway, see FPS website review, which is much cheaper.
There's a sweet spot and using the simple MHz metric as a golden ruler is weird at best... (Remember the old Pentium War).

Anyway, this is running circle. Let's see what AMD can do and decide for yourself if the chip meets your needs because you know, you can do more than gaming too lol.
 
Red Falcon said:
There are still a shit ton of programs out there that are strictly single-threaded, or are lightly-threaded where having that single boost core would greatly benefit them.
Audacity transcoding, Windows updates, clamd, database applications, many updater programs (both Windows and Linux), certain compression tools, etc.

Believe it or not, your OS still has many back-end programs that are strictly single-threaded. ;)
Click to expand...

im not disregardign what you say but i find it funny that those kind of people are focussed on the small gaing of 250mhz speed.
when most of these software can get 20 % performance boost but implement a better thread scheduling ( aka project mercury).
but hey that will require admitting that the CCX desing is hurting performance. and suddenly you are an Intel shill for understanding basic cpu design *sigh*

in short: properly assigned threads to avoid CCX jumping give way more performance boost to ryzen than this little issue
 
Gideon said:
At least boost clocks work better on Ryzen then your DLSS setting or the missing DLSS 2x on your Nvidia card, I remember you telling me that was the big reason you got the card. Yet here you are, shouldn't you be in a Nvidia forum raising hell about it? 200 MHz is unnoticeable except in a benchmark, but yeah that suddenly means I have 0 interest in overclocking. Back when I had 1 GHz processors or less then yeah that actually meant something. Take a 9900K at 4.8 then to 5.0 and tell me you actually notice a difference in how it runs. This whole thing is silly with false outrage, it's a boost clock and last time I saw that was not a guarantee you will hit them.
Click to expand...

I said I didn’t factor in RTX at all for my purchase but I was excited about DLSS2X, but I don’t count on mystery tech. I could go back before reviews even happened and quote myself if I really cared. I also posted very negatively in a recent DLSS thread.

Nice off topic post. Shows you don’t have a leg to stand on.

This thread has the most asinine mental gymnastics happening and then when called out they essientially attack the poster.

At the end of the day AMD messed up, there’s no denying it. Their numbers on the box are entirely misleading. Also, single threaded matters or Intel wouldn’t have a 20% lead in minimums at high Hz.

How much are our rigs of people that buy ~$500 processors? $3,000? $4,000? Lets save a $100 to gimp them (for gaming) on hopes AMD comes through.
 
Last edited:
Dayaks said:
I said I didn’t factor in RTX at all for my purchase but I was excited about DLSS2X, but I don’t count on mystery tech. I could go back before reviews even happened and quote myself if I really cared. I also posted very negatively in a recent DLSS thread.

Nice off topic post. Shows you don’t have a leg to stand on.

This thread has the most asinine mental gymnastics happening and then when called out they essientially attack the poster.

At the end of the day AMD messed up, there’s no denying it. Their numbers on the box are entirely misleading. Also, single threaded matters or Intel wouldn’t have a 20% lead in minimums at high Hz.

How much are our rigs of people that buy ~$500 processors? $3,000? $4,000? Lets save a $100 to gimp them (for gaming) on hopes AMD comes through.
Click to expand...

Or game at 1080P where you see the most gain using those CPUs ? As the resolution go up, the gain goes down but I'm sure you knew that.

Intel has a lead in gaming because of a combination of factors and frequency is one of those but I doubt it's single core boost, I mean comeon, unless you assigned your gaming tasks to that particular core, how can you be sure you benefit for it running faster ? For all you know your game might be running from the slower cores.
What SvenBent said is what makes the most sense in this thread.
 
Mode13 said:
Notable performance or not, its the precedence of this whole thing that's concerning. This is the same forum that went after Nvidia with razor sharp claws in recent memory but is now giving AMD a free pass for their own consumer unfriendly shenannigans.. 100mhz now sets the precedence for 500mhz next time (in fact, i must repeat that it IS ~400mhz or more for some 3900x users, counter arguments are lumping all users into the best case scenario to minimize the issue)

These chips are all binned, that's why we have a 3600x and 3800x. Look at silicon lotto results, virtually no 3700x will hit 3800x clocks. It should have been known that only one in 20 chips would see these boosts. They could have shipped a lower SKU or just printed reality on the box. "Up to" is nonsense, we're not advertising car insurance savings, put the clocks that the lowest denominator will reach on the box or split the sku.

I'm glad AMD is acknowledging it and trying to give us some boost back and I know they tried to push the chips to the absolute edge to get those competitive numbers, but in the end it's a bug or they just decided to be a bit unethical.
Click to expand...
Not a free pass, but people are acting like they stole their first born... No, they should have said a lower number, hopefully the bios fixes it, I just hope people don't have these expectations that it'll make a huge difference. I think it was a dumb move to try to explain that it's a theoretical max for a single core.... Nobody wants a theoretical # that they can't hit, but it isn't going to change much either.
 
MMitch said:
Or game at 1080P where you see the most gain using those CPUs ? As the resolution go up, the gain goes down but I'm sure you knew that.

Intel has a lead in gaming because of a combination of factors and frequency is one of those but I doubt it's single core boost, I mean comeon, unless you assigned your gaming tasks to that particular core, how can you be sure you benefit for it running faster ? For all you know your game might be running from the slower cores.
What SvenBent said is what makes the most sense in this thread.
Click to expand...

Or game in 1440p or VR... VR where you get motion sickness under 90Hz is not something I want to choose to do on purpose.

Yeah, if you’re at 4k or love IQ at 1440p this is basically a non-issue, or if you wait for OCing reviews (which usually take longer for in depth ones) you would also know. Status quo was you could get all core at or above max boost, with proper cooling, until AMD’s shenanigans.
 
Dayaks said:
Or game in 1440p or VR... VR where you get motion sickness under 90Hz is not something I want to choose to do on purpose.

Yeah, if you’re at 4k or love IQ at 1440p this is basically a non-issue, or if you wait for OCing reviews (which usually take longer for in depth ones) you would also know. Status quo was you could get all core at or above max boost, with proper cooling, until AMD’s shenanigans.
Click to expand...

You're stuck on 1 metric while the CPU still deliver the performance. If Bulldozer was running at 5.5GHz would it be better than Intel ? There's more to CPU than the megahurtz race.
I can understand the sentiment people have that they were cheated but let's rationalize and analyze the actual performance of the CPU and not this magic number which is MHz.

People are acting out of proportion here and I guess I could be viewed as someone that tries to brush it off or a fanboy if you want (I guess people that don't know me could say that... I currently own many machines and it's a mix.. depends what was the best bang for the buck for a specific task at the time of decision making). About the comment on Overclocking results, it has always been a lottery and to think we are entitled to XYZ results is a millennial way of thinking if you ask me. It's like when the police tolerates 10MPH over the limit (It's in KM/H here so I don't know what police in the US tolerates) and been mad when you get a ticket because you drove at 8MPH over the limit, well tuff luck those were the rules... you win you lose... same thing with OC.

Quick search on VR Review, you may want to reconsider what you typed... I may need to dig more before I can make any conclusion though. One sample is not much and I don't want to entertain this much more... you draw your conclusion.
https://www.vortez.net/articles_pages/amd_ryzen_7_3700x_review,18.html
 
MMitch said:
You're stuck on 1 metric while the CPU still deliver the performance. If Bulldozer was running at 5.5GHz would it be better than Intel ? There's more to CPU than the megahurtz race.
I can understand the sentiment people have that they were cheated but let's rationalize and analyze the actual performance of the CPU and not this magic number which is MHz.

People are acting out of proportion here and I guess I could be viewed as someone that tries to brush it off or a fanboy if you want (I guess people that don't know me could say that... I currently own many machines and it's a mix.. depends what was the best bang for the buck for a specific task at the time of decision making). About the comment on Overclocking results, it has always been a lottery and to think we are entitled to XYZ results is a millennial way of thinking if you ask me. It's like when the police tolerates 10MPH over the limit (It's in KM/H here so I don't know what police in the US tolerates) and been mad when you get a ticket because you drove at 8MPH over the limit, well tuff luck those were the rules... you win you lose... same thing with OC.

Quick search on VR Review, you may want to reconsider what you typed... I may need to dig more before I can make any conclusion though. One sample is not much and I don't want to entertain this much more... you draw your conclusion.
https://www.vortez.net/articles_pages/amd_ryzen_7_3700x_review,18.html
Click to expand...

VRmark is a canned benchmark and is garbage. Just go look at a normal benchmark with 1% or 0.1% minimums. It all applies.

A normal enthusiast knew what the IPC of the 2xxx series was, knew where the 3xxx series would be and also the information from AMD themselves. They had “4.75Ghz” in their marketing video around launch. I personally held off but was pretty hard core into buying a 3900x or 3950x. As more info came out I found out I wasted my time.

So I didn’t waste my money as we’ve seen deliberately misleading info from AMD in the past, waited for reviews, but it is annoying.
 
Last edited:
Mode13 said:
Notable performance or not, its the precedence of this whole thing that's concerning. This is the same forum that went after Nvidia with razor sharp claws in recent memory but is now giving AMD a free pass for their own consumer unfriendly shenannigans.. 100mhz now sets the precedence for 500mhz next time (in fact, i must repeat that it IS ~400mhz or more for some 3900x users, counter arguments are lumping all users into the best case scenario to minimize the issue)

These chips are all binned, that's why we have a 3600x and 3800x. Look at silicon lotto results, virtually no 3700x will hit 3800x clocks. It should have been known that only one in 20 chips would see these boosts. They could have shipped a lower SKU or just printed reality on the box. "Up to" is nonsense, we're not advertising car insurance savings, put the clocks that the lowest denominator will reach on the box or split the sku.

I'm glad AMD is acknowledging it and trying to give us some boost back and I know they tried to push the chips to the absolute edge to get those competitive numbers, but in the end it's a bug or they just decided to be a bit unethical.
Click to expand...

der8hauer found chips only could boost to 4025 MHz.
 
thebufenator said:
Oh look he's back.

Just curious, did der8hauer test them or was it from his survey?
Click to expand...

just the survey.. i believe all the chips he personally tested were only 25-50mhz below single threaded max boost.

i'd say anyone that was more than 150mhz below max single threaded boost probably had high ambient temps or they screwed up and ran the multithreaded test instead of the single core test. even so there were a ton of trolls though and people that had no business even filling out the survey selecting wrong cpu's which is why you saw some really stupid numbers for some of them being +200mhz above max boost which is literally impossible to do if you had your bios set to default.
 
sirmonkey1985 said:
just the survey.. i believe all the chips he personally tested were only 25-50mhz below single threaded max boost.

i'd say anyone that was more than 150mhz below max single threaded boost probably had high ambient temps or they screwed up and ran the multithreaded test instead of the single core test. even so there were a ton of trolls though and people that had no business even filling out the survey selecting wrong cpu's which is why you saw some really stupid numbers for some of them being +200mhz above max boost which is literally impossible to do if you had your bios set to default.
Click to expand...

Guess my 3700x only boosting to 4250 is just lying to me. Maybe my custom water isn't keeping the CPU around 50-60c, and that's lying to me as well. Or my component selection isn't "top tier" enough.

My shit is broke. Completely stock settings, custom settings, doesn't matter.
 
schmide said:
All that science under your belt and you only see the outliers. The Gaussian in me just died a little.

Even worse you phrase it to come off as all chips only boost to 4025mhz. True colors.
Click to expand...

48786DF9-F2B2-4013-941B-DD42E00F2F8E.png


Yeah, even derbauer said to ignore the results outside of 2 sigma. 4250 is still pretty low for a single core boost but like many said that could very well be a poor cooling solution. It’s interesting there is a cluster there though.
 
Last edited:
Hundreds of thousand of cpu's sold. derbauer polls less than 1000 and is only getting results from people who watch the fps counter in the upper right hand corner but don't play the game.
Non scientific poll is non scientific.

Anyone want to hazard a guess on what the next internet faux outrage will be about?
 
Status
Not open for further replies.
Back
Top