• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

Ryzen: Strictly technical

Page 76 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
Status
Not open for further replies.
mat9v said:
Well, they might have been able to do it like this: i.redditmedia.com/--EzP2t4YMf9TCYeibbG29tu3Ajfeje5QjrvCk8g1uI.png?w=572&s=d4d21987be37de7cfc9679f1aff4d5ea - but I do not know if RyzenMaster would correctly show it as 4500Mhz.
Click to expand...

I'll try it tomorrow.
 
I think someone showed non AVX around ~120W prime95 or aida in HWinfo at 4.3GHz for i7 8700K. Well maybe I forgot a bit when I saw some test, but there is where it starts to shine... you will hit 5GHz+ with only ~50W more. I think it was 4.3GHz stock vs 5.1GHz OC around 115-120W and 170-175W.


To much off topic on ryzen.
But I really like what intel did with 14nm++, instead of making soldered K processors they did...

Anyway I hope that 7nm will do better for AMD.
 
Last edited:
beginner99 said:
Am I missing something? isn't this the age old "TDP isn't power consumption" issue? If you use 140W, some of it is actually transferred to work and not all of it wasted as heat?
Click to expand...

AMD uses TDP as power consumption for CPU for a long time now. You are correct, TDP is something different. As you can see its very interesting how much more power R7 2700X needs for 50p more in CB R15.

@The Stilt testing:

No limit or 140W TDP = 1821p
Limit at 105W TDP = 1764p
When you look at his graph you can say that some CPus will do same power at 100MHz higher or other way around. Since @The Stilt went only for 4.1GHz I assume that was its maximum at safe voltages, which could mean that he got really bad CPU. Der8auer got 4.3GHz on all cores.
 
Last edited:
IRobot23 said:
No limit or 140W TDP = 1821p
Limit at 105W TDP = 1764p
When you look at his graph you can say that some CPu will do same power at 100MHz higher or other way around. Since @The Stilt went only for 4.1GHz I assume that was its maximum at safe voltages, which could mean that he got really bad CPU. Der8auer test he got 4.3GHz on all cores.
Click to expand...

Hm, so could this actually mean that with PBO and XFR2, everyone is subject to silicon lottery as worse chip will have lower limits for these features? So seeing quiet "big" differences in benches is to be expected?
 
beginner99 said:
Hm, so could this actually mean that with PBO and XFR2, everyone is subject to silicon lottery as worse chip will have lower limits for these features? So seeing quiet "big" differences in benches is to be expected?
Click to expand...

We will need to wait for @The Stilt so he can tell us how much can he reach with his CPU.
It's more about silicon lottery and probably AMD needed to go all in with 12nm, highest clocks with 8C and mass quantities. Probably same would happen with every other product on launch.
 
The Stilt said:
The 8700K is actually able to sustain the advertized all-core turbo frequency (4.3GHz) in most workloads, even when the power limits are configured as they should (95W/119W).
I tested this after 8700K was launched and basically everything else but highly 256-bit workloads maintained 4.3GHz at 95W.
Click to expand...
FYI Coffee Lake behaved the same when Z370 launched. My presumably 65W TDP i7 8700 installed in a MSI Z370 board had no power limit set in place, IIRC the max package power in Prime 95 was 130W+ and in CB 15 was 90W+.

It's all down to the MB manufacturer though, and I assume the same applies for Ryzen: on high performance boards they will forego power limitations by default to get bigger scores in benchmarks. Budget boards with lower spec power delivery will be configured to abide official TDP.
 
coercitiv said:
FYI Coffee Lake behaved the same when Z370 launched. My presumably 65W TDP i7 8700 installed in a MSI Z370 board had no power limit set in place, IIRC the max package power in Prime 95 was 130W+ and in CB 15 was 90W+.

It's all down to the MB manufacturer though, and I assume the same applies for Ryzen: on high performance boards they will forego power limitations by default to get bigger scores in benchmarks. Budget boards with lower spec power delivery will be configured to abide official TDP.
Click to expand...

That would mean default power consumption means almost nothing.

You'd have to go through everyones boards to see what the manufacturer implemented and for competitive reasons the high end boards will be chucking the power limits out the window for the extra edge.
 
IRobot23 said:
You said that R7 2700X is first one... this site shows every ryzen out of TDP.
Click to expand...

The chart clearly says they are measuring ATX 12V power. Since CPUs don't use 12V, CPU TDPs are calculated after the motherboard VRM efficiency losses. If you assume that most of the VRMs are somewhere between 80-90% efficient, suddenly a lot of the numbers align much more closely with their TDPs.
 
Gyronamics said:
You'd have to go through everyones boards to see what the manufacturer implemented and for competitive reasons the high end boards will be chucking the power limits out the window for the extra edge.
Click to expand...
The only premium boards one can still reasonably expect to play the spec sheet game are mITX boards, for obvious reasons 🙂
 
So I have seen some bclk oc, it's can boost 2/4 core to 4,5 ghz. Is that safe oc? By just letting motherboard control the voltage with offset.
 
CatMerc said:
Holy hell

My Gigabyte AX370 Gaming k7 with overdrive enabled claims it's pushing 1.7V through my CPU.
Click to expand...

I have to apply a negative offset voltage on my X370 board to get it to set voltages and boost properly. What is the agesa version of your Gigabyte board? I'm waiting for Asrock to put out a 1.0.0.2a BIOS still to see if that makes it work well without tweaking.
 
Look at the "CPU Core Voltage SVI2 TFN" (HWInfo) instead of what RyzenMaster or other applications are showing.
The CPU itself can never request higher than 1.550V voltage as 1.5500V is the highest VID the SVI2 interface supports.
Use of voltage offsets from the controller side is obviously possible, however the CPU itself cannot control those.
 
The Stilt said:
Look at the "CPU Core Voltage SVI2 TFN" (HWInfo) instead of what RyzenMaster or other applications are showing.
The CPU itself can never request higher than 1.550V voltage as 1.5500V is the highest VID the SVI2 interface supports.
Use of voltage offsets from the controller side is obviously possible, however the CPU itself cannot control those.
Click to expand...

That's what I've been looking at. My guess is that my board (X370 Taichi) provides a higher voltage than what the CPU actually requests on default settings causing safety limiters to trigger early, but I'm not sure how to verify. I was stuck in the 3.9x GHz boost range before applying the offset and now it sticks at 4.1 or just below, even in Prime95.
 
repoman0 said:
That's what I've been looking at. My guess is that my board (X370 Taichi) provides a higher voltage than what the CPU actually requests on default settings causing safety limiters to trigger early, but I'm not sure how to verify. I was stuck in the 3.9x GHz boost range before applying the offset and now it sticks at 4.1 or just below, even in Prime95.
Click to expand...

Could you post a screenshot of the newest HWInfo (showing the CPU Core SVI2 TFN value), while stressing the CPU with CPU-Z "stress CPU" option?

The board shouldn't be able to override the voltage, unless you've manually set the voltage to "fixed" mode.
Obviously in OC-Mode (CPU ratio manually changed) the voltage controller shuts down, but in normal mode it will always be enabled.
 
So even with power limits essentially lifted, you can still gain clockspeed by undervolting with an offset?

My board is already AGESA 1.0.0.2a.
 
repoman0 said:
Yes. I will prove it later.
Click to expand...

Actually, you're correct.

The cropping takes place only when the actual voltage is increased either through a voltage offset or load-line calibration.
I used -50mV offset and saw the voltage request to increase only by 6.25mV.

So technically there appears to be no Vmin tracking, while Vmax tracking is definitely there.
Obviously this is great for additional tinkering.
 
Status
Not open for further replies.

TRENDING THREADS

Back
Top