AMD Zen 2 (Ryzen 3000) - *** NO COMPETITOR HINTING ***

[hudson86]

9900K boosts one core to 5Ghz, 1800X boosts one core to 4.1Ghz. 3700X boosts one core to 4.4Ghz, what's the problem?

You are going out your way to avoid the point, so lets try this again.

9900k boosts to 5ghz , but ANY of the 8 cores can physically hit 5ghz
1800x boosts to 4.0ghz , but ANY of the 8 cores can physically hit 4.0ghz

3600X boosts to 4.4ghz, but ONLY 1 of the 6 cores can physically hit 4.4ghz

Or do you need this in pictures ?

 

[Skynet5]

Sarcasm I hope?

Lol. Yeah.

I find it pretty disgusting tbh. I get supply and demand. But it isn't a commodity.. It isn't crude oil. It isn't gold. Its a piece of silicon.

This is purely a price hike by the last person in the supply chain.

 

[Skynet5]

You are going out your way to avoid the point, so lets try this again.

9900k boosts to 5ghz , but ANY of the 8 cores can physically hit 5ghz
1800x boosts to 4.0ghz , but ANY of the 8 cores can physically hit 4.0ghz

3600X boosts to 4.4ghz, but ONLY 1 of the 6 cores can physically hit 4.4ghz

Or do you need this in pictures ?

Meow..

That's interesting though.. So only one core is designated to boost so high? Sounds like silicon lottery within silicon lottery.

 

[ljt]

It could be something to do with the latest BIOS/AGESA versions though too.

I had AGESA 1.0.0.1 on my X370 Taichi initially with my 3700X. HWInfo showed all 8 cores reached 4.4Ghz at some point during normal use

I updated BIOS to AGESA 1.0.0.3AB (as 1.0.0.1 had memory issues running above 3000Mhz) and now no cores boost to 4.4Ghz, the max I see on only 2 or 3 cores is 4.375Ghz, and about 4.325-4.350 on the rest.

This may be what toms hardware are seeing, so it could just need a tweaked AGESA to bring the boost clocks back to what they were on 1.0.0.1

 

[hudson86]

Sorry, just realised you have 3600 not 3600x , so am sure you can match your all core 4.2 with your advertised boost clock of 4.2

The higher rated CPUs seem much less likely to exhibit this behaviour and they've potentially binned good single cores for just these CPUS to give a better impression of boost clocks. Given this has been a popular story, boost clocks not being hit, the Toms analysis seems good.

Maybe its an early production issue, and future releases they'll have enough higher binned cores to allow all cores to hit max boost ? Still dodgy as ....

 

[humbug]

i think he's getting at the fact it's apparently only 1 specific core as opposed to any of the cores, singly, being able to hit the boost freq

I know he is, its fake outrage, its still boosting one core to the advertised speed, the algorithm will put the workload only using one thread, like the primary draw call thread onto that core, in operation the end result is no different to any other CPU. he and as always Toms Hardware are making a huge issues out of nothing.

He or Toms Hardware don't care that Intel advertise the 9900K as a 95 watt CPU when in fact its 160 Watt out of the box, but this, and many other nothing burger AMD crap is a huge issue to Toms Hardware.

 

[Skynet5]

I think having only one core being able to boost is a bit rubbish.

Isn't switching workloads from one core to another difficult?

I'm no hardware architect, but it sounds a bit restrictive.

 

[hudson86]

Fake outrage ? Nothing burger? You sound like a certain Orange person that gets upset when news is posted he doesnt like.

Its good info, especially for enthusiasts and owners wondering whats happening with Boost clocks not matching advertised speeds.

If you choose to ignore it or deem it irrelevant to your workload / usage , then fine, but its different behaviour to any previous generation of AMD or Intel.

 

[humbug]

I think having only one core being able to boost is a bit rubbish.

Isn't switching workloads from one core to another difficult?

I'm no hardware architect, but it sounds a bit restrictive.

No, as always it will put the most important work load on the fastest core, the end result is the same as everyother CPU, that's why this is a nothing story blown out of proportion.

 

[humbug]

No, as always it will put the most important work load on the fastest core, the end result is the same as everyother CPU, that's why this is a nothing story blown out of proportion.

If you read what Toms Hardware actually said you will see they pretty much said this themselves...

Technically, AMD's only specified boost clock applies to a single-threaded workload, which you could argue means AMD only has to deliver a single core capable of delivering the maximum frequency. But, if there are several slower cores that can only reach the base frequency, that would surely impact performance in various multi-threaded workloads. We hope that AMD provides more clarity in the days to come.

You can only off load a single threaded workload on to the singular fastest core, this is the same for all CPU's, their manufactured problem is that they don't boost ALL cores to this speed in "multithreaded workloads" all CPU's behave like this.

In games you have one worker thread, its the most important thread, its the one that bootlenecks, it will run on the fastest core, just like it does on the 9900K and 1800X.

 

[oweneades]

Just ran CB R15 Single core on my 3900x. Seems to hit 4.6Ghz on 4 cores and 4.5 on the other 2. I assume the software is only hitting one CCX as the load never hit the bottom 6 cores.



Not sure what Tom's were testing but the logic that the CPU can only hit 4.6Ghz on a literal single core doesn't seem to follow, at least not with my CPU / program load.

 

[hudson86]

No, as always it will put the most important work load on the fastest core, the end result is the same as everyother CPU, that's why this is a nothing story blown out of proportion.

And if your 4.4ghz core is busy and your workload is pointed at another core, lets say one thats only capable of 4.2 ? Again different behaviour to prev gen

 

[humbug]

And if your 4.4ghz core is busy and your workload is pointed at another core, lets say one thats only capable of 4.2 ? Again different behaviour to prev gen

Same with any CPU... Edit, it will move the other thread, it will prioritize the most important workload, like every other CPU.

https://forums.overclockers.co.uk/t...itor-hinting.18825852/page-1147#post-32910640

 

[triss]

Same with any CPU... Edit, it will move the other thread, it will prioritize the most important workload, like every other CPU.

https://forums.overclockers.co.uk/t...itor-hinting.18825852/page-1147#post-32910640

Well its not quite the same as any other cpu as the previous ryzens (excluding threadripper) and intels mainstream platforms every core is capable of hitting that boost clock at times.
I'm not saying its a huge issue atm but it is a little odd its not been mentioned by AMD until asked.

 

[chrcoluk]

You are going out your way to avoid the point, so lets try this again.

9900k boosts to 5ghz , but ANY of the 8 cores can physically hit 5ghz
1800x boosts to 4.0ghz , but ANY of the 8 cores can physically hit 4.0ghz

3600X boosts to 4.4ghz, but ONLY 1 of the 6 cores can physically hit 4.4ghz

Or do you need this in pictures ?

Dont know about 1000 series but interestingly my 2600X only uses one of two cores for its single core clock.

It only ever uses core 1 and 4 out of 6 for the single core boost. I had noticed whenever I tried an all core clock matching 4.25ghz, it would hard lock even with voltages bigger than XFR 4.25ghz, so I started watching XFR and noticed it only ever uses those 2 cores. So 0,1,2,3,4,5.

I dont have an issue with this tho and am glad AMD push these chips to the absolute maximum at shipping state.

My theory is the cores get binned during manufacturing, and if cores are marked as unable to hit max clock, they get disabled in XFR for the fast speed bin. So it probably varies from chip to chip and not relevant to the generation of ryzen. The 3000 series chips have the highest clocks so is logical they will be harder to get more cores able to hit those clocks. Also it can use both core 1 and 4 at same time at 4.25ghz, so its not limited to single core in XFR.

 

[DragonQ]

You are going out your way to avoid the point, so lets try this again.

9900k boosts to 5ghz , but ANY of the 8 cores can physically hit 5ghz
1800x boosts to 4.0ghz , but ANY of the 8 cores can physically hit 4.0ghz

3600X boosts to 4.4ghz, but ONLY 1 of the 6 cores can physically hit 4.4ghz

Or do you need this in pictures ?

I don't see any technical/legal issues if this is indeed the case, as long as the OS understands which core is the "best" and always uses that core first. I'm not convinced Windows is clever enough to do this though.

From an enthusiast point of view, it kinda sucks because it means all-core overclocks will suffer.

It could be something to do with the latest BIOS/AGESA versions though too.

I had AGESA 1.0.0.1 on my X370 Taichi initially with my 3700X. HWInfo showed all 8 cores reached 4.4Ghz at some point during normal use

I updated BIOS to AGESA 1.0.0.3AB (as 1.0.0.1 had memory issues running above 3000Mhz) and now no cores boost to 4.4Ghz, the max I see on only 2 or 3 cores is 4.375Ghz, and about 4.325-4.350 on the rest.

This may be what toms hardware are seeing, so it could just need a tweaked AGESA to bring the boost clocks back to what they were on 1.0.0.1

Same, I get lower boost clocks with 1.0.0.3AB compared to 1.0.0.1. It didn't help memory speeds for me though, so worse all round.

 

[Martini1991]

Clocks in ryzen 3XXX are a little crapper than expected. Don't think anyone can argue with that lol.

Seems if you're really lucky you'll hit 4.4GHZ all core on a 3900X.

 

[chrcoluk]

Just ran CB R15 Single core on my 3900x. Seems to hit 4.6Ghz on 4 cores and 4.5 on the other 2. I assume the software is only hitting one CCX as the load never hit the bottom 6 cores.



Not sure what Tom's were testing but the logic that the CPU can only hit 4.6Ghz on a literal single core doesn't seem to follow, at least not with my CPU / program load.

that score seems low, 4.4ghz ryzen 3700X chips match that score.

 

[oweneades]

No idea, its running on my test bench with the Wraith Max cooling it (making a right racket as well), 3000C16 RAM from my wifes machine with no tuning beyond XMP, no bios changes and running on a boot drive that is from my old X99 machine (not reformatted) with no manual shutdown of background processes. Only running as is to test if the beta drivers allow Destiny 2 to work (they do).

Just checked, KitGuru scored 209 so seems in line given the above.

 

[Tnaik44]

Hello guys, i m new here but been following this thread.
I upgraded 3 days ago from my i7 6700k to :
ryzen 3900x
asus strix x570 f-gaming
kraken x62 with noctua fans on it
ADATA xpg spectrix d60g 2c8gb 3200cl16.

My pbo boost all cores to around 4100 and single core to around 4275. No matter what i tinker in the bios its the same, what is weird is i did a manual OC and was able to get all 4350mhz @1.28v fully stable with aida64 stability test for many hours and temps were 68-69c.
Its clearly a bios or software issue with pbo since thw cpu can go higher while using lower voltage.
And please if u can help, i m not really an expert with ryzen and ddr4, is the kit i got good for 3900x , if not can anyone tell me what should i get if money is no issue, just the best i can get to complement ryzen 3900x.

Thank you a lot , this thread been very helpful .