Ryzen 5 3600 turbo speeds

[Toothless]

I just popped in to see what condition my condition was in...

After eyeing up a 5800X i might just try some experimental frying of my 3600 on a B450 board.
Who's with me for a 3600 @5GHz on all cores?

I've got a few bucks left if you wanna give that 3600 a home.

Honestly burnt silicon doesn't smell good to the point of I'd suggest don't try it.

 

[nguyen]

I am nowhere near afraid of degrading my 5600 as I am with my 3600. But I still wouldn't push more than 1.3v on it for anything long term. I am ok with 1.25v though. It has seen 1.45v by my hand for 4900Mhz briefly, but that's the highest I've benched. All core load on it with 1.3v @ 4700 with linpack and I can barely keep temps in check (under 85), but not only that, it is sucking 140w+. My 3600 pretty much stops reacting in a stable fashion to all core voltage @ 1.337v for 4500MHz. And linpack at that speed and voltage is about the same on the 5600X, very hard to tame and imo borderline dangerous, if not fully dangerous.

Everyone knows the crap cores come from the outer wafer. Everyone knows crap cores don't scale like good cores. No one can say with certainty what an acceptable all core voltage is with 7nm. In my opinion 1.25v is safe for any load under the sun. Whatever your FIT voltage is, you shouldn't go to far past it. The problem is further exasperated if your cooling solution isn't up to the task.

This was the only trial run I ran with the 3600 (stock + PBO)

4.1Ghz/1.37Vcore reduce the power consumption and temperature, so I guess that would mean lower current as well. Temperature is kept in check at all time with the Kraken X63 AIO, too bad I didn't record with 4.1ghz/1.37Vcore

 

[Hemmingstamp]

I've got a few bucks left if you wanna give that 3600 a home.

Honestly burnt silicon doesn't smell good to the point of I'd suggest don't try it.

You're in the US. The import taxes alone would be more than what the 3600 is worth.

Just thought I'd lighten the mood around here

 

[Deleted member 202104]

I just popped in to see what condition my condition was in...

After eyeing up a 5800X i might just try some experimental frying of my 3600 on a B450 board.
Who's with me for a 3600 @5GHz on all cores?

I love the smell of melting silicon in the morning.

 

[WhitetailAni]

I love the smell of melting silicon in the morning.

Combined with the beeping of your UPS because the power went out.

 

[damric]

If you find this helpful, I'll re-run the test with default motherboard settings and/or PBO.

Ryzen 3600 @4200MHz
X470 Taichi
32GB 3800CL16

Tested with AVX ON. Graph set to 1 second intervals and about a 15 minute test of CPU/FPU.

Edit: here's a better pic with the load numbers.

 

[freeagent]

I guess mine wasn't as hot as I remembered.. but that could have been December.. was pretty cold outside by then



Moving to 4x8 looks like things warmed up a bit. Hard to say because my fan configuration was in the experimental stage, and was constantly changing. Or maybe the furnace came on..

 

[MancLad]

No, it's best to leave him to his delusions of 1.37v being safe.

I'm not delusional here and there is no need to get funny about it. Like I said on my first post, I'm new to OC and just wanted some 'friendly' advice.

 

[Zach_01]

Found some old (2~3 months ago) tests with my bronze sample 3600 with PBO Enabled.
Anyone care to see stock voltage and speed during CB-R20 and P95(AVX)?

Here you go

P95 AVX 128-128KB

1.29V (avg), 3.915GHz and 111+W PPT (avg)

------------------------------------------------------

CB R20 MT

1.35V avg, 3.995GHz and 94+W PPT

-------------------------------------------------------------------

Keep in mind that those voltages and speeds are under the specific (each test) CPU temps. If temp was lower, voltage and speed would be higher and vice versa. Also same voltages and speeds are under the specific type of loads. Any other type of load would result (or required) different voltages and speeds.
If I use 1.35V for a static OC I must set speed no more than 4.0GHz, keep temps at 60C max, and never ever load an AVX load on the CPU.

 

[MancLad]

That's single core voltage. Even 1.5 is safe for single core loads. You had 1.4v for all loads before, including all core loads, which is very dangerous.

For all core loads, let the CPU boost accordingly, it will select a voltage it thinks is safe. 1.325 is not safe for all core on Zen 2, I'd stay at 1.25 max and even then I'd rather have the CPU run on stock. You're no smarter than the CPU, once you set your own voltage you disable all safety features like the voltage fitness regulator and the fate of the CPU is in your hands, so it's best to keep it at stock. If you want more performance, get a good cooler, a board with solid VRMs and enable PBO.

Thanks for the reply. I do have a good cooler and board.

Thanks for the reply. I do have a good cooler and board

Keep hearing this 'PBO' for overclocking. I've gone back into the BIOS and set to D.O.C.P for my RAM and just set the rest to auto!

 

[Toothless]

You're in the US. The import taxes alone would be more than what the 3600 is worth.

Just thought I'd lighten the mood around here

Bet. Still would do it.

 

[Hemmingstamp]

Bet. Still would do it.

If I do sell I'll give you first refusal

 

[damric]

Found some old (2~3 months ago) tests with my bronze sample 3600 with PBO Enabled.
Anyone care to see stock voltage and speed during CB-R20 and P95(AVX)?

Here you go

P95 AVX 128-128KB

1.29V (avg), 3.915GHz and 111+W PPT (avg)

View attachment 199207

------------------------------------------------------

CB R20 MT

1.35V avg, 3.995GHz and 94+W PPT

View attachment 199209

-------------------------------------------------------------------

Keep in mind that those voltages and speeds are under the specific (each test) CPU temps. If temp was lower, voltage and speed would be higher and vice versa. Also same voltages and speeds are under the specific type of loads. Any other type of load would result (or required) different voltages and speeds.
If I use 1.35V for a static OC I must set speed no more than 4.0GHz, keep temps at 60C max, and never ever load an AVX load on the CPU.

Unlucky sample if you need that much voltage for such low clock.

My Cinebench R23

 

[Zach_01]

Unlucky sample if you need that much voltage for such low clock.

My Cinebench R23

View attachment 199223

If you like comparisons you'll have to run CB-R20 and observe "CPU Core Voltage SVI2 TFN" and speed on HWiNFO with "Snapshot CPU Polling" Enabled from main settings.

 

[damric]

If you like comparisons you'll have to run CB-R20 and observe "CPU Core Voltage SVI2 TFN" and speed on HWiNFO with "Snapshot CPU Polling" Enabled from main settings.

lol why. R20 is weaker than R23.

 

[Toothless]

lol why. R20 is weaker than R23.

Because, he ran R20, as said in his post.

 

[Zach_01]

lol why. R20 is weaker than R23.

For sake of comparison. You can't compare different loads on a different binned CPU, even if its the same SKU and talk about voltages and speeds between them. I don't have R23. So I'm asking you, if you wouldn't mind running it (R20) like I did.

If you do so
Start R20, wait for CPU values to go to 100% (like 4~5sec), then reset HWiNFO values and take a screenshot at 1:10~1:15

 

[WhitetailAni]

If you'd like, I can record my R20 and/or R23 results and recorded voltages once I get home (in ~2 hours). It's currently on stock clocks, but RAM is 4x8GB 3333 C16-16-16-32 tRC 48.

 

[Zach_01]

If you'd like, I can record my R20 and/or R23 results and recorded voltages once I get home (in ~2 hours). It's currently on stock clocks, but RAM is 4x8GB 3333 C16-16-16-32 tRC 48.

RAM speed doesn't really matter for CB as long as it can keep the CPU fed up. Anything between 3200~3800 DRAM speed is practically the same for this benchmark. And even if it does make difference, for sure we are not comparing scores here, just CPU core voltages and speeds.
So yeah, please do so (R20 preferably). Just try to do it like I described at post #167.

EDIT:

And PBO Enabled (not Auto) for direct comparison as well.

 

[WhitetailAni]

Alright, results:

Spoiler

I've underlined what I think are important values.

No idea how to interpret them, though, so other people can explain it (to me as well).

 

[Lindatje]

Ryzen 3600 here. From the start 2019 on 1,39v and still running very stable and smooth.

 

[biffzinker]

Ryzen 3600 here. From the start 2019 on 1,39v and still running very stable and smooth.

I have a dead Ryzen 3600 from that year that stopped POSTing mid 2020. The early abuse with a manual multi/voltage, and PBO enhanced may of contributed to it's death. Working fine one day, and then the next nope, CPU LED lit on the board. Same board is fine with a 3800X, and the temp Ryzen 3 1200.

 

[Mussels]

Funny that I get lower core temperature with static 4.1ghz/1.37Vcore on my 3600 by 10C compare to stock clocks/voltages with PBO, so I guess I might as well be extending the life of my 3600, but not to the point that is last longer than its usefulness period (5-7 years).
Well if you treasure PC hardware that much why not just use 1.1Vcore then , why bother with 1.25VCore at all?

It's not temps that kill them, its amperage. Thats straight from AMD engineers.

 

[tabascosauz]

I have a dead Ryzen 3600 from that year that stopped POSTing mid 2020. The early abuse with a manual multi/voltage, and PBO enhanced may of contributed to it's death. Working fine one day, and then the next nope, CPU LED lit on the board. Same board is fine with a 3800X, and the temp Ryzen 3 1200.

I only tested my 3700X at sub-1.25V, but I did run it for a few months PBO with LLC. Turbo LLC on that IR35201 was unbelievably strong. And then I ran a aggressive PBO profile for a long time, with uncapped PPT, 83A EDC and 2X scalar. That 2X scalar is no joke, all-core Vcore damn close to 1.4V.

AMD imposed specific current/clock restrictions on AVX-type Vcore with 1.0.0.6/ComboV2, but some B450 boards never implemented those R20 and P95 limits (~Q1/Q2 2020) even when stock, PBO disabled. My 3 B550 boards, even with PBO on, were never able to replicate the ~4900-5000pt R20 stock, no PBO scores on my B450I Aorus Pro Wifi.

Did all that have a part to play in my 3700X's shenanigans later in life? Who knows.

BZ says that it takes a lot of effort to degrade any Matisse chip. Which makes sense, since I did do a LOT of stupid shit over 18 months. But iirc he mostly used his 8700K/9900K and modded Z390 Gene for the OC stuff through 2019 and early 2020, so I don't think his 3700X actually got much use before he made the degradation video earlier this eyar.

And then there's my no-fucks-given 4.4 @ 1.45V one week frenzy on my 4650G just to get that 9891 score in the R23 thread we don't talk about that except to say that it will only do 4.0 now lol

 

[mclaren85]

I'm really happy to see someone else also share my views (which is I firmly believe right). Those zen2 architecture requires excessive voltage already (design error maybe?) and that will certainly degrade the chip sooner or later. I would not push the chip more if I were you.