Ryzen benchmarking and overclocking results

[infrared]

Wow, that's a screaming fast R5 (comment meant for @Johan45, but applies to that hwbot screenshot above too, holy crap!)

I may have bought a new 1800x from silicon lottery, quite enjoying this chip:

Joint first place on HWBot for 1800x on custom h2o. I tried for 4.25ghz but it couldn't do it at 1.506v, since i didn't plan to go over 1.5 anyway i called it a day. Pretty happy though. voltage in cpuz is accurate in this case. 1.48v LL4 set in bios.

Edit: currently crunching away at 4.1GHz 1.39v, which is +150mhz over last one.

Edit #2: Kinda interesting comparison, this cinebench score puts this in between a bunch of 5960X's at 4.8GHz.

 

[cdawall]

Its benchmarking, not 24/7...grow a pair or go home.

TPU cant seem to discern between the two very well.

This always happens. Jesus my soc voltage made people cry, but I had the ram running 3600-3800

 

[Kanan]

Again, I'm not against overclocking and benchmarking for fun, but it's no relevant information regarding "better Ryzen yields" and "higher frequencies". It's also way too early for that anyway. At least wait 6 months for that to chime in, if it's not limited by architecture anyway.

People here seem to have a hard time to understand the simplest of things. A single CPU doing what, 4.25 GHz, at high voltage, because of silicon lottery luck, is no fucking evidence on "better Ryzen yield" and "higher frequencies". Now this is the last time I'm gonna explain it, after that I'm in ignore mode and laughing.

 

[Super XP]

CPUz does report the voltage a bit higher than it really is. Either way, I rather lower CPU speed a bit to sustain stability and keep the voltage at bay. Too high and you might kill the CPU. Perhaps.

 

[phanbuey]

Wow, that's a screaming fast R5 (comment meant for @Johan45, but applies to that hwbot screenshot above too, holy crap!)

I may have bought a new 1800x from silicon lottery, quite enjoying this chip:

Joint first place on HWBot for 1800x on custom h2o. I tried for 4.25ghz but it couldn't do it at 1.506v, since i didn't plan to go over 1.5 anyway i called it a day. Pretty happy though.

Edit: currently crunching away at 4.1GHz 1.39v, which is +150mhz over last one.

Edit #2: Kinda interesting comparison, this cinebench score puts this in between a bunch of 5960X's at 4.8GHz.

View attachment 87034

they have definitely improved yields

 

[Kanan]

they have definitely improved yields

Proof? You're talking nonsense. It's way too early for that.

 

[Johan45]

People here seem to have a hard time to understand the simplest of things. A single CPU doing what, 4.25 GHz, at high voltage, because of silicon lottery luck, is no fucking evidence on "better Ryzen yield" and "higher frequencies". Now this is the last time I'm gonna explain it, after that I'm in ignore mode and laughing.

It's not a silicon lottery chip, bought through Canadacomputers. You're right though it's nothing to do with better yields I just got lucky with a better than average Ryzen.
This is also an overclocking and benchmarking thread so I wasn't under the impression that stability was a must.

Wow, that's a screaming fast R5 (comment meant for @Johan45, but applies to that hwbot screenshot above too, holy crap!)

I may have bought a new 1800x from silicon lottery, quite enjoying this chip:

Joint first place on HWBot for 1800x on custom h2o. I tried for 4.25ghz but it couldn't do it at 1.506v, since i didn't plan to go over 1.5 anyway i called it a day. Pretty happy though.

Edit: currently crunching away at 4.1GHz 1.39v, which is +150mhz over last one.

Edit #2: Kinda interesting comparison, this cinebench score puts this in between a bunch of 5960X's at 4.8GHz.

Spoiler

View attachment 87034

That HWBot link was my 1700x which still works exactly as it did prior to freezing it. @infrared Nice chip, much better than most. Can you get it to do 3800 Mem ? I managed once on my 1700x but couldn't get windows to start again after, just can't seem to stabilize it.

 

[ratirt]

Like I said in my post 1.42V at the CPU. CPUz always reports way too high on these. There's also a bit of droop going on. That voltage is fine and no it's not 24/7 but not crazy either. Besides when done right it doesn't hurt the CPU and it is a fine candidate for LN2. That's when it gets crazy muahahaha. I did learn something else If I wanted to run 3600 ram I had to drop 50 MHz for the same V_Core so it wasn't a total loss.
Now this is crazy http://hwbot.org/submission/3509166_

If the ryzens could sustain this OC with a reasonable voltage on air or at least water. That would be something

 

[EarthDog]

Yeah a 200 mhz overclock over xfr! Lets get excited!!!!!

 

[the54thvoid]

My chip is air temp stable at 3.9Ghz and 1.45v. Prior to that it was unstable at 1.43v. It would crash in BF1 during loading screens. It's now been solid gaming for 11 days and counting.

So... Time to try 3.95?

Edit: @EarthDog -it's the TPU way to get every bit out of the chip is it not? An increase is still an increase

 

[infrared]

It's not a silicon lottery chip, bought through Canadacomputers. You're right though it's nothing to do with better yields I just got lucky with a better than average Ryzen.
This is also an overclocking and benchmarking thread so I wasn't under the impression that stability was a must.


That HWBot link was my 1700x which still works exactly as it did prior to freezing it. @infrared Nice chip, much better than most. Can you get it to do 3800 Mem ? I managed once on my 1700x but couldn't get windows to start again after, just can't seem to stabilize it.

Lucky git!

I'm memtest86+ stable (well, 1 pass of test 7) at 3700mhz with 18-18-18-38 timings on the 2666mhz divider with 1.46V VDimm and 1.25V SoC, trying to nudge it up but not having any luck booting at 3725 atm.
Edit: I give up, tried different dividers, and SoC up to 1.275 and vdimm up to 1.475v but no luck. 3700mhz seems to be the limit of the IMC/inf. fabric on this one.
Now... Who do i buy some LN2 gear off? haha Joking for now, but this needs to happen!

@EarthDog / @Johan45 any chance you can hit me up next time you're doing an ln2 sesh?

My chip is air temp stable at 3.9Ghz and 1.45v. Prior to that it was unstable at 1.43v. It would crash in BF1 during loading screens. It's now been solid gaming for 11 days and counting.

So... Time to try 3.95?

Edit: @EarthDog -it's the TPU way to get every bit out of the chip is it not? An increase is still an increase

Dunno, going above 1.40V you get quickly diminishing returns, and these things churn out a lot more heat at 1.45. I wouldn't want to go any higher for a daily setting, even on water. You could deffo get to 3.95ghz for short bench runs or something though

 

[EarthDog]

My chip is air temp stable at 3.9Ghz and 1.45v. Prior to that it was unstable at 1.43v. It would crash in BF1 during loading screens. It's now been solid gaming for 11 days and counting.

So... Time to try 3.95?

Edit: @EarthDog -it's the TPU way to get every bit out of the chip is it not? An increase is still an increase

This morning, and yesterday, TPU and a few long time users have really really disappointed me... oh well, its just me... so who cares, but damn, the dumb in this place has grown exponentially lately.

 

[64K]

This morning, and yesterday, TPU and a few long time users have really really disappointed me... oh well, its just me... so who cares, but damn, the dumb in this place has grown exponentially lately.

 

[Tatty_Two]

This morning, and yesterday, TPU and a few long time users have really really disappointed me... oh well, its just me... so who cares, but damn, the dumb in this place has grown exponentially lately.

Stay positive, focus on the other 354 days of the year

 

[Johan45]

Lucky git!

I'm memtest86+ stable (well, 1 pass of test 7) at 3700mhz with 18-18-18-38 timings on the 2666mhz divider with 1.46V VDimm and 1.25V SoC, trying to nudge it up but not having any luck booting at 3725 atm.
Edit: I give up, tried different dividers, and SoC up to 1.275 and vdimm up to 1.475v but no luck. 3700mhz seems to be the limit of the IMC/inf. fabric on this one.
Now... Who do i buy some LN2 gear off? haha Joking for now, but this needs to happen!

@EarthDog / @Johan45 any chance you can hit me up next time you're doing an ln2 sesh?


Dunno, going above 1.40V you get quickly diminishing returns, and these things churn out a lot more heat at 1.45. I wouldn't want to go any higher for a daily setting, even on water. You could deffo get to 3.95ghz for short bench runs or something though

That's kind of where mine stopped. 3744 IIRC was the best I could manage at decent timings, never found I needed the SOC that high though. Don't think I have been over 1.1 TBH. If you haven't try PLL to 2.0V and VDDP or VDPP can't get that straight ATM at 1.2v I found both those helped at higher frequencies. This was bench stable don't know about actual stability but good enough for X265 4K which is a tough ass bench
Oh and I set the VTT to 0.85-0.9v

 

[infrared]

That's kind of where mine stopped. 3744 IIRC was the best I could manage at decent timings, never found I needed the SOC that high though. Don't think I have been over 1.1 TBH. If you haven't try PLL to 2.0V and VDDP or VDPP can't get that straight ATM at 1.2v I found both those helped at higher frequencies. This was bench stable don't know about actual stability but good enough for X265 4K which is a tough ass bench
Oh and I set the VTT to 0.85-0.9v

Nicely done, yeah x265 bench at 4K is pretty brutal. I was having a stab at that earlier but was always a couple of fps down on the others at 4.2ghz so i need to figure out why, maybe just closing down loads of services, or that silly bias option in the bios. I can't believe your 3744mhz aida bench was at CL14, v.nice.

I'll give some of those settings a try, thanks for giving me some hints The SOC was just me clutching at straws tbh, it didn't actually get me any further than 1.15V iirc. VDDP was set to 1.02V so I'll try bumping that up next time. PLL was already at 2.0. I had the ram at 1.46v so set VTT to .73v based on the default 1/2 vdimm, I'll try going higher on that too.

What voltage do you find the mem works well at? I seem to have a weird thing going on where 1.43v is good, 1.44-1.45v appears to make things worse, then getting past that 1.46v helped out with the 3700mhz, above that I can't boot anyway so don't really know. I've heard samsung b-die likes high voltage but not sure where the limits are, I don't really want to hurt the nice trident z's

 

[Kanan]

I never said I'm only interested in practical 24/7/365 overclocks. That said overclock the heck out of it if you want, I wouldn't care and I'm still curious about it.

@the54thvoid honestly I would've sold that chip or gave it back to get a higher clocking one. I mean you paid the 100 bucks extra for the 1700X - so you deserve more than that imo.

 

[infrared]

Edit: sorry @Kanan , replying to the wrong person.

 

[Johan45]

What voltage do you find the mem works well at? I seem to have a weird thing going on where 1.43v is good, 1.44-1.45v appears to make things worse, then getting past that 1.46v helped out with the 3700mhz, above that I can't boot anyway so don't really know. I've heard samsung b-die likes high voltage but not sure where the limits are, I don't really want to hurt the nice trident z's

For those timings/speed 1.6V dropping to 3600 same timings 1.5-1.55V

 

[ratirt]

Yeah a 200 mhz overclock over xfr! Lets get excited!!!!!

I see on the link over 5.2 that's what I was referring to bro. 2 bad you missed it Maybe other time you will notice the difference

 

[the54thvoid]

This morning, and yesterday, TPU and a few long time users have really really disappointed me... oh well, its just me... so who cares, but damn, the dumb in this place has grown exponentially lately.

Yeah, saw that. We've never been hardcore like OCN with our outlook on overclocking.

Still...My own project was get 3.9Ghz on air on all cores and I got that. Happy enough.

 

[cdawall]

This morning, and yesterday, TPU and a few long time users have really really disappointed me... oh well, its just me... so who cares, but damn, the dumb in this place has grown exponentially lately.

It comes in spurts. Maybe it is just their time of month?

 

[Tatty_Two]

Yeah, saw that. We've never been hardcore like OCN with our outlook on overclocking.

Still...My own project was get 3.9Ghz on air on all cores and I got that. Happy enough.

We used to be better/more interested, there was a period in my first 3 or 4 years where there was quite a following here, both extreme and just enthusiast, damn I literally used to buy every new CPU and board combo that was released, cost me a fortune, I do think that with the more modern CPU's and platforms they have taken away some of the challenges and perhaps that is in part to blame for the lack of ambition, certainly at the extreme level, that's why it is quite nice to see Ryzen offering a few fresh challenges.

 

[Johan45]

We used to be better/more interested, there was a period in my first 3 or 4 years where there was quite a following here, both extreme and just enthusiast, damn I literally used to buy every new CPU and board combo that was released, cost me a fortune, I do think that with the more modern CPU's and platforms they have taken away some of the challenges and perhaps that is in part to blame for the lack of ambition, certainly at the extreme level, that's why it is quite nice to see Ryzen offering a few fresh challenges.

It most certainly is and I've been enjoying it honestly. Some new/different voltages to play with and try to understand their functions. Just a shame they run out of gas so quickly.
This is just for reference on the Ryzen voltage reporting with CPUz. I have highlighted the "actual" CPU voltage in HWinfo64 to demonstrate the disparity. This was an LLC setting 03 on the CHVI with it on auto it was dropping below 1.39V and causing some stability issues.

 

[Kanan]

It most certainly is and I've been enjoying it honestly. Some new/different voltages to play with and try to understand their functions. Just a shame they run out of gas so quickly.
This is just for reference on the Ryzen voltage reporting with CPUz. I have highlighted the "actual" CPU voltage in HWinfo64 to demonstrate the disparity. This was an LLC setting 03 on the CHVI with it on auto it was dropping below 1.39V and causing some stability issues.

View attachment 87051

Is this the newest CPU Z you got installed? I thought I saw it report 2x4 Cores as in accounting for CCX'es.

I think Intel killed a lot of fun off by limiting overclocking to K cpus. Those Skylake parts that were able to overclock despite not being K on Z boards weren't very prominent in the overclocking scene because most bought the K cpus anyway.