Did you get a shiny new Ryzen 3000 CPU?

[Chomiq]

@TheLostSwede thanks. Yeah, the massive spread of idle voltage from 0.2V to 1.5V is a major concern for me. When you stress test with CPB off, does it actually hit 4.5GHz or does it just say that and actually top out at 3.9GHz in HWInfo?

Just another issue with saving money on a board not X570, I guess. Gigabyte does seem to release a lot of BIOSes for my board to keep it up to date with AGESA, but this problem clearly demonstrates that something's amiss.



Unfortunately, that's a bit like burying my head in the sand. With CPB on, it's still feeding a 1.4V spike into the CPU regardless of what the fan setting is. Right now the fan reaction time is delayed but still enough to keep the temps from reaching 70C. Reducing the polling rate would just allow the CPU to roast. The fans would still kick in on longer spikes anyways.

CPB off clearly demonstrates that it does know how to behave itself, it just chooses not too on CPB.

Nm then, I now noticed that you're using M1v5.

 

[biffzinker]

@TheLostSwede thanks for the heads up, it's getting late and my patience is wearing thin with AMD.

The only way out I can think of is to manually re-adjust the fan settings in the BIOS, and stop looking at the voltage spike during boost in anything else but Ryzen Master. The other option is to find the maximum all cores heavy load overclock to get away from the 1.5V, also allows you to determine the highest voltage. The cores will still go into power saving state.

Doesn't do squat in terms of thermals at idle for me though, but under full load the CPU obviously doesn't get as hot.

Sounds like the same as me disabling XFR, and PBO on A MSI motherboard. Last time I did it limits the CPU to just running at base clock speed.

 

[tabascosauz]

Why is the spread an issue? Some of the cores are in deep sleep, so they go down to 0.2V, this happens normally for me now on the latest UEFI on my board.
It was an issue on early UEFI's where these ultra low Voltages weren't reported properly, but AMD fixed that in AGESA 1.0.0.3ABB and the new drivers.

And no, I can't boost at all with CPB disabled.

The spread is an issue because it hits idle voltages of 0.2-1.5V. Load hits 1.3-1.4V which is still too high compared to the Vcore setting reported in the BIOS at 1.225V.

What's intriguing is that the BIOS is giving it 1.225V to work with up to 3.6GHz, but Ryzen is only deciding to use 1.01V even at that speed, which is only 0.1V above usual idle. That "missing" voltage gives me an idea.

As for idle, it's improved for me as well. 3700X idles in a very narrow range between 37-40C and strictly 0.9V with occasional dips to 0.2V. With CPB on, it's anywhere from 31-55C and 0.2/0.9/1.3/1.5V.

The only way out I can think of is to manually re-adjust the fan settings in the BIOS, and stop looking at the voltage spike during boost in anything else but Ryzen Master. The other option is to find the maximum all cores heavy load overclock to get away from the 1.5V, also allows you to determine the highest voltage. The cores will still go into power saving state.

Sounds like the same as me disabling XFR, and PBO on A MSI motherboard. Last time I did it limits the CPU to just running at base clock speed.

Maybe CPB is what Gigabyte calls PBO? It's a long shot.

The fact that it might be about PBO lit up a faint light bulb. I'm thinking what I might be able to do is do away with boost entirely and manually adjust base clock and Vcore up in increments to see how close to max boost clock I can get without torpedoing thermals, using that 1.225V I've been given. After all, with CPB off, it works like a normal processor entirely within the range of 0 to Base Clock, which I should be able to adjust manually in BIOS. Which, coincidentally, is what I had to do with my 4790K.

Begrudgingly, it makes me appreciate Intel's experience tuning Turbo Boost for different TDPs over the years.

 

[Chomiq]

The spread is an issue because it hits idle voltages of 0.2-1.5V. Load hits 1.3-1.4V which is still too high compared to the Vcore setting reported in the BIOS at 1.225V.

What's intriguing is that the BIOS is giving it 1.225V to work with up to 3.6GHz, but Ryzen is only deciding to use 1.01V even at that speed, which is only 0.1V above usual idle. That "missing" voltage gives me an idea.

As for idle, it's improved for me as well. 3700X idles in a very narrow range between 37-40C and strictly 0.9V with occasional dips to 0.2V. With CPB on, it's anywhere from 31-55C and 0.2/0.9/1.3/1.5V.



Maybe CPB is what Gigabyte calls PBO? It's a long shot.

The fact that it might be about PBO lit up a faint light bulb. I'm thinking what I might be able to do is do away with boost entirely and manually adjust base clock and Vcore up in increments to see how close to max boost clock I can get without torpedoing thermals, using that 1.225V I've been given. After all, with CPB off, it works like a normal processor entirely within the range of 0 to Base Clock, which I should be able to adjust manually in BIOS. Which, coincidentally, is what I had to do with my 4790K.

Begrudgingly, it makes me appreciate Intel's experience tuning Turbo Boost for different TDPs over the years.

PBO is PBO, it's in AMD Overclocking section in BIOS and is enabled by default.

Aorus Master x570 Perfect UEFI Settings

Setup> advanced mode. Tweaker Enable xmp. CPU vcore -> normal CPU clock control 100.01 Settings AMD overclocking > Ddr and infinity fabric frequency/timings > if dividers > set this to half your memory clock. E.g. 3600 mem = 1800 in this menu. Note that I don’t recommend 3600 – 3533 or a...

forum.level1techs.com

 

[Adam Krazispeed]

I got me a tasty new Ryzen 3700x from microcenter (sharonville Ohio) newegg and every where else (except EBAY ) were out of stock of every cpu except the 3600/3600X, was like WTF>>!!!

my first 3700X died tho after about 2 hours, used in my Asrock fatal1ty X370 Professional gaming AM4 MB (1st Gen) flashed bios to Agesa combo 1.0.0.1 )WTF sucj a low Version 1.0.0.3AB was out for all other MBs except mine. (now tho 1.0.0.3ABB is out on my x370 mb but removes PCI-e Gen4 support(MFkers)

but took my bad 3700x back an exchanged it for another , this new one works way better can OC all Cores to 4.35Ghz ALL CORES/Threads.. actually get better ST than running it astock???
i won silicon lottery this time for sure lol

but just went up yesterday and bought my Gigabyte X570 Aorus master ($359.99) Microcenter shaonville ohio

my systems

#1 X570 - 3700X 8c/16t Gigabyte X570 Aorus Master

#2 Current X399 Gigabyte aorus Gaming 7 R TR 1900X 8c/16t aug 2017

#3 Old 1st Gen ASROCK fatal1ty X370 Gaming Professional (Ryzen 7 1800X) Early Adopter, ordered on march 2nd 2017)
tried out R5 2600X on 2nd gen soon to be and itx build once i get an itx mb fo my R7-1800X or R5-2600X old cpus..

**next build is gonna be an ITX gaming pc build using older parts, need an itx MB x370/x470, but need to find a good case small but can hold say up to a 1080 Ti or 2080 Ti for gpu upgradability.. and can support GPU Hybrid cooled cards also? rad support

 

[TheLostSwede]

The spread is an issue because it hits idle voltages of 0.2-1.5V. Load hits 1.3-1.4V which is still too high compared to the Vcore setting reported in the BIOS at 1.225V.

What's intriguing is that the BIOS is giving it 1.225V to work with up to 3.6GHz, but Ryzen is only deciding to use 1.01V even at that speed, which is only 0.1V above usual idle. That "missing" voltage gives me an idea.

As for idle, it's improved for me as well. 3700X idles in a very narrow range between 37-40C and strictly 0.9V with occasional dips to 0.2V. With CPB on, it's anywhere from 31-55C and 0.2/0.9/1.3/1.5V.

Maybe CPB is what Gigabyte calls PBO? It's a long shot.

The fact that it might be about PBO lit up a faint light bulb. I'm thinking what I might be able to do is do away with boost entirely and manually adjust base clock and Vcore up in increments to see how close to max boost clock I can get without torpedoing thermals, using that 1.225V I've been given. After all, with CPB off, it works like a normal processor entirely within the range of 0 to Base Clock, which I should be able to adjust manually in BIOS. Which, coincidentally, is what I had to do with my 4790K.

Begrudgingly, it makes me appreciate Intel's experience tuning Turbo Boost for different TDPs over the years.

No, CPB is something entirely different. PBO is in a different part of the UEFI and on the X570 board you have to agree to a disclaimer before you can even access it.
This is with CPB disabled, but PBO enabled (well, auto). You can see PBO is enbabled because of the high values for PPT, TDC and EDC.

As for the Voltages, AMD has already said that it's normal to see peaks of 1.5V, so not sure what you're worrying about.

 

[londiste]

**next build is gonna be an ITX gaming pc build using older parts, need an itx MB x370/x470, but need to find a good case small but can hold say up to a 1080 Ti or 2080 Ti for gpu upgradability.. and can support GPU Hybrid cooled cards also? rad support

X370/X470 is quite pointless for gaming system. The only obvious benefit over B350/B450 is Crossfire which isn't a thing for ITX either way

 

[tabascosauz]

No, CPB is something entirely different. PBO is in a different part of the UEFI and on the X570 board you have to agree to a disclaimer before you can even access it.

As for the Voltages, AMD has already said that it's normal to see peaks of 1.5V, so not sure what you're worrying about.

I can't remember, is the trend to smaller process complementary with a trend to higher voltages, or is it an inverse relationship? In any case, a good number of boards (mostly Asus, but not all) seem to be dumping obscene amounts of Vcore into their chips. I'd like to see if I can get past 4GHz without using any more than 1.225V. Maybe I've just been a 22nm guy, not used to seeing such high voltages.

I can't seem to see much of a difference with Default and PBO settings in Ryzen Master, because as long as CPB is on, everything behaves like PBO.

X370/X470 is quite pointless for gaming system. The only obvious benefit over B350/B450 is Crossfire which isn't a thing for ITX either way

Not to mention most X470 ITX boards were literally identical to their B450 siblings except for what was under the PCH heatsink, yet considerably more expensive. Unless you're trying to snag one used.

 

[TheLostSwede]

Well, Precision Boost is always enabled, the PBO version of it, shouldn't be, but seems to be on Auto, i.e. enabled on Gigabyte boards. It also seems to make ZERO difference once you ned up using AGESA 1.0.0.3ABB. Prior to that, it didn't do anything...

 

[Vario]

Switch over NVMe to the first slot:
View attachment 130026

Nice bump in write performance.

 

[tabascosauz]

Well, Precision Boost is always enabled, the PBO version of it, shouldn't be, but seems to be on Auto, i.e. enabled on Gigabyte boards. It also seems to make ZERO difference once you ned up using AGESA 1.0.0.3ABB. Prior to that, it didn't do anything...

Also, what's up with the Vcore settings in GB's UEFI? It's "auto" by default, but most people suggest "normal". There's no pop-up window. Do I have to type in "Normal" manually into the box? That's a real bruh moment, Gigabyte.

Maybe that'll help calm things down.

 

[TheLostSwede]

Also, what's up with the Vcore settings in GB's UEFI? It's "auto" by default, but most people suggest "normal". There's no pop-up window. Do I have to type in "Normal" manually into the box? That's a real bruh moment, Gigabyte.

Maybe that'll help calm things down.

Hit enter on any option to get a list of options, or hit N to change from Auto to Normal.

 

[Nordic]

What are people's thoughts on AMD releasing bios that will lower boost clocks a bit in favor of stability? There was some comment about it earlier but not much discussion.

 

[TheLostSwede]

What are people's thoughts on AMD releasing bios that will lower boost clocks a bit in favor of stability? There was some comment about it earlier but not much discussion.

Considering I've had ZERO stability issues, it sounds odd. It's only a rumour so far, so let's wait and see...

 

[EarthDog]

Same. No stability issues across several boards and bios... but still not reaching single core boost either.

 

[Nordic]

I finally found a build where someone listed their temps with a 3900x and full custom water cooling. They had temps of 60c at full load. It is absolutely hard to cool these chips. The density is real.

 

[tabascosauz]

Okay, day 2 of testing.

As @TheLostSwede predicted, PBO on or off makes no difference. It seems that CPB acts like an MCE of sorts, except that it's a really spiteful breed of MCE that won't let you have your boost clocks if you turn it off. CPB is looking like an archetypal yandere at this point.

Here's what I have so far:

- CPB on, all clocks and voltages Auto ("1.225V"): 3700X boosts like a madman causing temps / volts / PWM to spike randomly.
- CPB on, all clocks and voltages Auto ("1.225V"), PBO off: no difference, if anything, more of a madman. Will peak at 65C on desktop under literally no load.

- CPB off, all clocks and voltages Auto ("1.225V"), (3.6GHz): 3700X is hard limited to 3.6GHz.

• OCCT Large Data and Prime95 Blend are stable. No errors.
• CPU maxes out at 100% load at 1.050V Vcore and 0.994V VID for all cores - literally nothing.
• Does not exceed 53C in OCCT Large Data 15min.
• Obviously, benchmark scores are disappointing.

- CPB off, all voltages Normal (lie-o-meter "1.225V"), manually set to 4.0GHz: 3700X is hard limited to 4.0GHz.

• A one-off graphical artifact occurred on the UAC secure desktop, while launching HWInfo; unassuming desktop background looked like something out of a horror movie.
• OCCT Large Data detects up to 7 errors within 5 seconds of initiating test, automatically stops the test ("crashes"). Consistently reproducible result.
• Prime95 results in a BSOD (KERNEL AUTO BOOST LOCK ACQUISITION WITH RAISED IRQL) within 0.5 seconds of beginning Blend.
• Vcore does not exceed 1.05V.

- CPB off, all voltages Auto (lie-o-meter "1.225V"), manually set to 4.0GHz: 3700X is hard limited to 4.0GHz.

• Identical results to above, minus graphical artifact.
• Combined with above run, litters WHEA Event Log (errors) with dozens and dozens of entries within half an hour.
• Vcore does not exceed 1.1V.

- CPB off, Vcore manually set to 1.3V, manually set to 4.2GHz: 3700X is hard limited to 4.2GHz.

• Idle is a bit warmer than at 4.0GHz, and with light load will move into 40s and 50s. But fan ramp-up and slowing down are progressive and smooth.
• OCCT Large Data is not stable. No errors until almost 4min30sec into test, when BSOD (KERNEL AUTO BOOST LOCK ACQUISITION WITH RAISED IRQL).
• Same 1.3V Vcore, down to 1.25V at max droop. VID still 1.1V.
• Full load OCCT temps reach 73C.

- For shits and giggles, CPB off, Vcore manual 1.3V, 4.4GHz:

• Cannot boot Windows.

EDIT:

- CPB off, Vcore manually set to 1.3V, manually set to 4.1GHz: 3700X is hard limited to 4.1GHz.

• Cannot initialize Prime95 on one of the worker threads. Fatal error. Other workers do just fine, but that's a no-no.

- CPB off, Vcore manually set to 1.3V, manually set to 4.0GHz: 3700X is hard limited to 4.0GHz.

• Idle is not spiky, but is warmer than 3.6GHz, dipping into the 40s occasionally.
• OCCT Large Data is stable. No errors.
• Prime95 Small FFT is stable in the short term (~19 minute test). Smallest FFT is not feasible not because of stability, but because it will break 90C.
• In OCCT, full load hovers around 1.3V mark Vcore, with up to 0.05V of droop. VID is a fair bit lower at 1.1V for all cores.
• Full load temps are consistently around 70C in OCCT.
• Prime95 Small reaches a thermal equilibrium of sorts around 87C.

It looks the 3700X is happy at 4.0GHz for now, with manual input of 1.3 volts. Problem is, the volts don't come down at idle because it counts as "manual overclocking" when I set the clocks and volts. HWinfo, CPU-Z report 1.27-1.3V at idle. Ryzen Master is confused and reports continuous 1.099999V.

 

[TheLostSwede]

I think you worry too much about things that are "normal" for Ryzen 3000.
The problem is that AMD hasn't been great at communicating what is normal.
My Voltages fluctuates like crazy and so does the temperatures, however, if you've read any of the limited feedback from AMD, this is how the CPUs are designed.
The Ryzen 3000 thermal sensors react quite differently to previous generations of CPUs and they seem to report the hottest temperature in the chip and the least bit of load, will cause temperatures to spike, but they also drop back just as quickly. This is "strange" behaviour compared to past generations of CPUs and it's something AMD could've communicated much better.
Yes, I was also wondering about all of this, but from what I have managed to find, it's all the new "normal".
What I cant' complain about is stability, as despite all of the weirdness, my system is 100% stable.

 

[Space Lynx]

Does anyone know if the free 3 month game pass for an AMD purchase expires? I won't be doing my final build for awhile still.

 

[Chomiq]

Does anyone know if the free 3 month game pass for an AMD purchase expires? I won't be doing my final build for awhile still.

*Offer available through participating retailers for eligible purchases made July 1, 2019 through March 10, 2020 or when supply of coupon codes is exhausted, whichever occurs first. Void where prohibited. Full offer terms at www.amdrewards.com/terms. Game Pass for PC: Over 100 PC Games available starting August 2019. Gears 5 available fall 2019. Game Pass code must be redeemed by June 30, 2020. Limit one promotional 3-month subscription per Microsoft account over a 12-month period. Requires the Xbox (beta) app and Windows 10 (with updates). Age restrictions and system requirements apply. Game catalog varies over time. Learn more at Xbox.com/gamepass.

 

[Space Lynx]

cool, so I got until June of next year. thanks Chromiq.

 

[Sybaris_Caesar]

cool, so I got until June of next year. thanks Chromiq.

... or when supply of coupon codes is exhausted

hehehe. Iirc RX 500 game bundle codes also ended early.

 

[Chomiq]

^ this

I've entered my code on Sunday, still waiting for it to activate.

 

[Space Lynx]

^ this

I've entered my code on Sunday, still waiting for it to activate.

ouch... oh well, it's only $4.99 a month for the Game Pass anyway, I already subscribe to it because it's so cheap and a great value.

I just downloaded CrossCode on there, a great game! Honestly if Microsoft can keep Game Pass this cheap... $7.99 in 2021, and like $9.99 in 2023... I'd gladly stay subscribed for several years, lot of great games already and more added weekly. It's an amazing deal.

I will pass on Ubisoft's though, $15 a month for theirs... no thanks... and skipping Origin as well because they have too many tiers and it gives me a headache finding which game I need for what tier so I said screw it, plus cost too much imo what you get, and Origin is the only client that ever crashes on me.

I'm surprised how well Microsoft has executed the Game Pass, very unlike them to do something right, sadly price will probably skyrocket once it comes out of beta, and I will unsub then. /shrug

 

[TheLostSwede]

Does anyone know if the free 3 month game pass for an AMD purchase expires? I won't be doing my final build for awhile still.

The Xbox (beta) app is super buggy. Worst "game launcher" I've used to date.