Ryzen 2600x OC

[hollyh88]

10 minutes ago, Tristerin said:

Example - this is how my runs look when OCing (this is a pic of mine from a few months ago) I.E. even Cine recognizes the OC.

 

 

 

i dont know man. all i do know is that yours has 1 more option. where is single core run with mine? maybe the cinebench i have is wrong? like i said i dont know. your the cinebench master here the speeds that hwm provides are however correct. they show ingame too. and with ryzen master. 

[Tristerin]

Just now, Bacon soup said:

@Tristerin I have a 2600 and a MSI B450 Mortar Titanium. If I use AMD's Ryzen tool, then cinebench does not see my overclock. Cinebench R20 is also able to crash my CPU where others can't. When Cinebench fails to see my cpu freq, I know it has overclocked because it does not crash on stock settings and I can see the heat rising. This has only been my first night of overclocking. I will try again in two weeks when my cooler, and other important parts, arrive.

That could explain the changes across the software, appreciate that insight.  

 

However it doesn't explain cooler runs from R15 going to R20 - R20 is way heavier on AVX instructions and will run much hotter than R15.  And in both cases he ran the same temp (71c) or a lower temp (68c) on R20 compared to R15.

 

Also the 1.5vCore...I just cant comprehend (still waiting to understand this, need someone smarter then myself for sure) how a Hyper 212 can accomplish this based on my extensive self taught knowledge about cooling an OCd CPU.

[hollyh88]

5 minutes ago, Bacon soup said:

@Tristerin I have a 2600 and a MSI B450 Mortar Titanium. If I use AMD's Ryzen tool, then cinebench does not see my overclock. Cinebench R20 is also able to crash my CPU where others can't. When Cinebench fails to see my cpu freq, I know it has overclocked because it does not crash on stock settings and I can see the heat rising. This has only been my first night of overclocking. I will try again in two weeks when my cooler, and other important parts, arrive.

cinebench r20 does more instructions so it is tougher on overclocks and can actually notice more overclocking failures due to that. thats why r15 can do an overclock flawlessly but as soon as you go and do it on r20 it crashes.

[hollyh88]

1 minute ago, Tristerin said:

That could explain the changes across the software, appreciate that insight.  

 

However it doesn't explain cooler runs from R15 going to R20 - R20 is way heavier on AVX instructions and will run much hotter than R15.  And in both cases he ran the same temp (71c) or a lower temp (68c) on R20 compared to R15.

 

Also the 1.5vCore...I just cant comprehend (still waiting to understand this, need someone smarter then myself for sure) how a Hyper 212 can accomplish this based on my extensive self taught knowledge about cooling an OCd CPU.

well that report might have been a spike. all in all i dont know. i can say that the case i have has brought down my temps significantly. used to have a single front fan intake sharkoon case. a t3 orso. compared to the Cooler Master MasterBox MB511 RGB   i have now with huuuge airflow intake. 

[Bacon soup]

2 minutes ago, Tristerin said:

However it doesn't explain cooler runs from R15 going to R20 - R20 is way heavier on AVX instructions and will run much hotter than R15.  And in both cases he ran the same temp (71c) or a lower temp (68c) on R20 compared to R15.

 

69C is my load temp on stock settings and stock cooler. Ambient is prob less than 10C

[Tristerin]

9 minutes ago, hollyh88 said:

i dont know man. all i do know is that yours has 1 more option. where is single core run with mine? maybe the cinebench i have is wrong? like i said i dont know. your the cinebench master here the speeds that hwm provides are however correct. they show ingame too. and with ryzen master. 

I should take a step back and ensure I espouse this - sorry for seeming so aggressive - Im after the knowledge here.  And to me, none of this is possible on a Hyper 212 Black.  Are you using Delta fans at 10k RPM?  Are you using cold air induction from outside or air conditioner?  There are factors in play here that science has to be able to explain and frankly - the surface area and TDP of the Hyper 212 Black (150w TDP) just doesn't cut it.

 

5 minutes ago, Bacon soup said:

69C is my load temp on stock settings and stock cooler. Ambient is prob less than 10C

Stock and at 10c (50F) and you are sitting at the same temps as a 1.5vCore OC (on stock cooler which is a good cooler but a little less than a Black 212) at higher ambient temps

OPs temps make no sense. 

 

I mean, I have the parts.  I just don't have this exact CPU.  I can do all this same testing (again) on my 160w air cooler (PCCooler brand), 240mm Corsair H100i AIO, and on my loop.  Guess what - only the loop could dissipate the heat from my similar setup and voltages though I do have more cores.  Maybe that's the difference?  I still don't understand lol.

[Bacon soup]

@Tristerin When I say less than 10C, it could have been zero. Right now its 7am, the sun has come up, and the internet says it is 11C. I don't have a thermometer. 

[Zando_]

1 minute ago, Tristerin said:

I should take a step back and ensure I espouse this - sorry for seeming so aggressive - Im after the knowledge here.  And to me, none of this is possible on a Hyper 212 Black.  Are you using Delta fans at 10k RPM?  Are you using cold air induction from outside or air conditioner?  There are factors in play here that science has to be able to explain and frankly - the surface area and TDP of the Hyper 212 Black (150w TDP) just doesn't cut it.

 

If you are doing OCs in a 50 degree F room that will help but eventually the stock cooler will heat soak.

The 2600 is incredibly easy too cool at stock, it's OCed where it gets hotter. At 1.35v (pulling up to 1.39v or so due to LLC on the B450-F), the 2600 I was tinkering with yesterday hit 80-85C, running a 4Ghz OC (this is in Prime95 blend, slightly higher in small FFT and slightly lower in CB20) on the stock cooler from an R5 1600 (due to the copper core I used that instead of the 2600 stock cooler when I built the rig). At 1.45v it would be a lot harder to keep under control, if you gave it 1.5v then yeah, absolutely bamboozled as to how a 212 keeps it at any reasonably safe temp. On the stock R7 2700X cooler (IIRC it's about on par with the 212) my 2700X would eat up to 1.5v under boost, then slap the 85C temp limit and throttle slightly to stay at/below that. That was boosting a single core, maybe two. An all core OC on 6c/12t at 1.5v should be a volcano. 

[Tristerin]

1 minute ago, Bacon soup said:

@Tristerin When I say less than 10C, it could have been zero. Right now its 7am, the sun has come up, and the internet says it is 11C. I don't have a thermometer. 

Ambient is the temperature of the room or area you are in.  I.E. my thermostat is set at 20c (70f) at my house, so my ambient at all times I quote at 20c

 

If you are doing these OCs outside to get those gainz - that's fine, the good overclockers do that lol - but needs to be noted when achieving an OC as its not "normal"

[Bacon soup]

17 minutes ago, Tristerin said:

Ambient is the temperature of the room or area you are in.  I.E. my thermostat is set at 20c (70f) at my house, so my ambient at all times I quote at 20c

 

we dont have fancy thermostats in nz. We just have windows that we open and close manually.

[Zando_]

1 minute ago, Bacon soup said:

we dont have fancy thermostats in nz. We just have windows that we open and close manually.

As the son of a kiwi mum I can confirm this 

[hollyh88]

im back with new screenshots  
i set back the oc to a 4.1 on all core on 1.4v

this is what i got:
in cb15 it hovered between 67/70.
in cb20 it hovered between 73/75. 76 was a 1 sec thing as it speeded up. 

oh and yeah dont look the dram stuff  like i said i used xmp profile 2 and this is stable without issues. so thats something i dont want to mess with for now.

 

[Zando_]

5 minutes ago, hollyh88 said:

im back with new screenshots  
i set back the oc to a 4.1 on all core on 1.4v

this is what i got:
in cb15 it hovered between 67/70.
in cb20 it hovered between 73/75. 76 was a 1 sec thing as it speeded up. 

oh and yeah dont look the dram stuff  like i said i used xmp profile 2 and this is stable without issues. so thats something i dont want to mess with for now.

Aha, those numbers are making a lot more sense. I kept an R5 1600 cool at 1.4v with a 120mm AIO, IIRC the 2600 isn't any hotter so those are much more reasonable numbers. 

[Tristerin]

2 minutes ago, hollyh88 said:

im back with new screenshots  
i set back the oc to a 4.1 on all core on 1.4v

this is what i got:
in cb15 it hovered between 67/70.
in cb20 it hovered between 73/75. 76 was a 1 sec thing as it speeded up. 


 

-snip-

 

 

Need help understanding how the scores didn't raise across the OC in R15, and how the score in R20 is negligible on a half a gigahertz increase.  That should be around a 50-65 point increase per 100mhz based from what I understand of R20 against the CPU.  The point of this program is to benchmark the gains and Im just not seeing them.  

 

Is this normal?  (asking the general audience)

[hollyh88]

2 minutes ago, Zando Bob said:

Aha, those numbers are making a lot more sense. I kept an R5 1600 cool at 1.4v with a 120mm AIO, IIRC the 2600 isn't any hotter so those are much more reasonable numbers. 

maybe my pc is just a bit eh weird? or i have a lose gun cpu that changes how it likes  its actually a highly advanced AI lmao. would love to see the 4.2 work on cinebench r20 but it just keeeps crashing. while gaming and cinebench r 15 goes without issues. 

[hollyh88]

4 minutes ago, Tristerin said:

 

Need help understanding how the scores didn't raise across the OC in R15, and how the score in R20 is negligible on a half a gigahertz increase.  That should be around a 50-65 point increase per 100mhz based from what I understand of R20 against the CPU.  The point of this program is to benchmark the gains and Im just not seeing them.  

 

Is this normal?  (asking the general audience)

probably has to do with outdated r15. it says windows 8 while im on windows 10 lol. 
 

[Zando_]

1 minute ago, hollyh88 said:

maybe my pc is just a bit eh weird? or i have a lose gun cpu that changes how it likes  its actually a highly advanced AI lmao. would love to see the 4.2 work on cinebench r20 but it just keeeps crashing. while gaming and cinebench r 15 goes without issues. 

Cinebench R20 uses AVX instruction sets (IIRC that's what they're called) on CPUs that support it. They're much, much harder to run and put the hurt on CPUs a lot more than other stuff. Most newer platforms let you set an AVX offset in the BIOS for this very reason, usually like 300Mhz by default. Meaning your CPU will run your full OC, but as soon as it hits something using AVX it'll downclock 300Mhz for stability, then hop back up when it isn't using AVX. 

You can ignore benches/stress tests that have AVX if everything you use normally runs fine, but as soon as you hit an AVX workload it's BSOD time, so I wouldn't recommend it. 

[Tristerin]

4 minutes ago, hollyh88 said:

probably has to do with outdated r15. it says windows 8 while im on windows 10 lol. 
 

R20 is where most of my experience lays (and no master by any means just seeking to understand) and with 1/3rd the cores but similar IPC (similar enough) between my R7 1700 and R5 2600X - you should be seeing gains in R20 of 1/3rd less (theoretically) than I get when OCing.  Above you can see I get around 130 points per 100mhz.  You should be getting a third less than that (theoretically)

 

Im going to just have to stick my 160w TDP air cooler on my OC and see how it handles it to get a better understanding of this when I clean the loop out.

[hollyh88]

27 minutes ago, Tristerin said:

R20 is where most of my experience lays (and no master by any means just seeking to understand) and with 1/3rd the cores but similar IPC (similar enough) between my R7 1700 and R5 2600X - you should be seeing gains in R20 of 1/3rd less (theoretically) than I get when OCing.  Above you can see I get around 130 points per 100mhz.  You should be getting a third less than that (theoretically)

 

Im going to just have to stick my 160w TDP air cooler on my OC and see how it handles it to get a better understanding of this when I clean the loop out.

i honestly dont know what to tell you i dont oc to get points on it.  i guess if you mix an air cooler with a case that has airflow it will perform alright. sure i can get a nice water aio cooler that may downgrade temps a bit but as linus already pointed out.. air coolers can even beat an aio or do equally as good. so it just really depends on what you put under it. ill just stick with this 4.1 since its at the 1.4v "safe" zone and doesnt crash r20. 

 

 

34 minutes ago, Zando Bob said:

Cinebench R20 uses AVX instruction sets (IIRC that's what they're called) on CPUs that support it. They're much, much harder to run and put the hurt on CPUs a lot more than other stuff.

which is why ill stick to my current oc. so it can even handle those workloads. maybe ill try vinding the perfect voltage for 4.2 another time. maybe some people that have the same cpu could bring their achievements to this post. 

 

[Tristerin]

12 minutes ago, hollyh88 said:

i honestly dont know what to tell you i dont oc to get points on it.  i guess if you mix an air cooler with a case that has airflow it will perform alright. sure i can get a nice water aio cooler that may downgrade temps a bit but as linus already pointed out.. air coolers can even beat an aio or do equally as good. so it just really depends on what you put under it. ill just stick with this 4.1 since its at the 1.4v "safe" zone and doesnt crash r20. 

 

 

which is why ill stick to my current oc. so it can even handle those workloads. maybe ill try vinding the perfect voltage for 4.2 another time. maybe some people that have the same cpu could bring their achievements to this post. 

 

Eh you don't have to chase points, it just gives values - computing doesn't have an opinion and its why I love it.

 

I have a Cooler Master HAF XM - definition of High Air Flow so I understand thermal dynamics as well as any shadetree PC guy whose always DIYing and trying new configs.  My other test case is a Cougar MX330-X which is a mesh case as well.  I also have a nice AIO (H100i GTX) that cant contain a 1.4vCore so again, still sounds fishy being that your air cooler isn't even close to high end.  So yes a high end air cooler, competing against a high end AIO makes sense, but that's middle of the road or budget at best for air coolers.

 

What still baffles me is the completely inaccurate reporting across the Apps.  If it were me, as I enjoy proving myself out, is pull the CMOS and reset your BIOS and start fresh.  1.) you may find a more stable OC 2.) that way your values being reported are accurate.

 

The thing is, you are in the top 10 world wide for Air CPU OC's on that chip in a CineR20 run if what your reporting is accurate (still trying to figure out how a Black Edition is taking 1.4-1.5vCore) so unlikely to get someone with input (I am also in the top 10 WW for my R7 1700 for CineR20).  EDIT - pretty sure a 1.5vCore pass would require LN2 and you show a 1.5vCore pass...

 

Im not jealous of your achievements, Im baffled by the heat sinks capabilities.  It still doesn't add up to me.  Maybe that's because Ive been only OCing 8 core CPUs in the time Ive been an OCer and the difference could be dropping 2 cores of heat coming off the IHS.  I don't know, but still doesn't add up to me.  I wonder if I can park 4 threads in Windows and see what my temps are after a CineR20 run compared to 16 threads going.

[hollyh88]

2 minutes ago, Tristerin said:

Eh you don't have to chase points, it just gives values - computing doesn't have an opinion and its why I love it.

 

I have a Cooler Master HAF XM - definition of High Air Flow so I understand thermal dynamics as well as any shadetree PC guy whose always DIYing and trying new configs.  My other test case is a Cougar MX330-X which is a mesh case as well.  I also have a nice AIO (H100i GTX) that cant contain a 1.4vCore so again, still sounds fishy being that your air cooler isn't even close to high end.  So yes a high end air cooler, competing against a high end AIO makes sense, but that's middle of the road or budget at best for air coolers.

 

What still baffles me is the completely inaccurate reporting across the Apps.  If it were me, as I enjoy proving myself out, is pull the CMOS and reset your BIOS and start fresh.  1.) you may find a more stable OC 2.) that way your values being reported are accurate.

 

The thing is, you are in the top 10 world wide for Air CPU OC's on that chip in a CineR20 run if what your reporting is accurate (still trying to figure out how a Black Edition is taking 1.4-1.5vCore) so unlikely to get someone with input (I am also in the top 10 WW for my R7 1700 for CineR20).  

 

Im not jealous of your achievements, Im baffled by the heat sinks capabilities.  It still doesn't add up to me.  Maybe that's because Ive been only OCing 8 core CPUs in the time Ive been an OCer and the difference could be dropping 2 cores of heat coming off the IHS.  I don't know, but still doesn't add up to me.  I wonder if I can park 4 threads in Windows and see what my temps are after a CineR20 run compared to 16 threads going.

well i know that there were improvements between the 1000 series and 2000 series. especially on what it can take. 
I cant say i know why something indeed like a hyper 212 black edition can take it. 

dont get me wrong my old case heated it up, this case has 3 front fans 1 back fan pulling air out. there is a mesh ontop and at the psu below. there is nothing blocking the fan of the cpu. 

I really dont wanne reset the bios via cmos. the cmos is kinda eh shitty to get to on this mobo. i would have to bring my case downstairs take out gpu take out cables and cpu cooler and then i can do it safely. a big hassle. i did however simply reset the bios within the bios. it has made hwm a bit more accurate, maybe i just have an outdated version and i never noticed. idk. 

i really think the 1.5v was just a spike. although it did report it on all the cores with pbo. maybe i actually have won the lottery and my cpu can handle way more than others would?

the 212 black edition does keep it around 73 on all cores. wich is still hot im sure an all water loop with pump would take it down further. but im impressed. i do have to note that i was thinking about getting the noctua cooler now that its finally not in hidious brown. but i will do that when i upgrade my cpu. as clearly i dont need it at all now. 

[Tristerin]

Just now, hollyh88 said:

well i know that there were improvements between the 1000 series and 2000 series. especially on what it can take. 
I cant say i know why something indeed like a hyper 212 black edition can take it. 

dont get me wrong my old case heated it up, this case has 3 front fans 1 back fan pulling air out. there is a mesh ontop and at the psu below. there is nothing blocking the fan of the cpu. 

I really dont wanne reset the bios via cmos. the cmos is kinda eh shitty to get to on this mobo. i would have to bring my case downstairs take out gpu take out cables and cpu cooler and then i can do it safely. a big hassle. i did however simply reset the bios within the bios. it has made hwm a bit more accurate, maybe i just have an outdated version and i never noticed. idk. 

i really think the 1.5v was just a spike. although it did report it on all the cores with pbo. maybe i actually have won the lottery and my cpu can handle way more than others would?

the 212 black edition does keep it around 73 on all cores. wich is still hot im sure an all water loop with pump would take it down further. but im impressed. i do have to note that i was thinking about getting the noctua cooler now that its finally not in hidious brown. but i will do that when i upgrade my cpu. as clearly i dont need it at all now. 

I have 2 golden sample R7 1700's (my other can also do a 3.8ghz all core on stock voltages, but I don't have a cooler - 120mm aio on it right now and 240mm AIO cant contain voltage increases well enough - to mitigate the 4.2 Im chasing on it as well (have the loop just haven't installed it yet, same exact loop on my RR2 in sig configuration)

 

But whats getting me, and the reason Im balking at the numbers (heat) is that they don't do "golden sample" heat sinks, which is what Im questioning.  If you apply the voltage, the heat has to go somewhere or the socket will catch fire (basically)

[hollyh88]

Gaming runs fine too on that oc

6 minutes ago, Tristerin said:

But whats getting me, and the reason Im balking at the numbers (heat) is that they don't do "golden sample" heat sinks, which is what Im questioning.  If you apply the voltage, the heat has to go somewhere or the socket will catch fire (basically)

well i do know that some mobos handle heat better. and even some coolers perform better than others of the same brand and well making. 
also.. if you wanne know the thermal paste i have on it it's this one: https://www.coolblue.nl/product/767423/titan-ttg-g30030-koelpasta.html.

[Tristerin]

16 minutes ago, hollyh88 said:

Gaming runs fine too on that oc

well i do know that some mobos handle heat better. and even some coolers perform better than others of the same brand and well making. 
also.. if you wanne know the thermal paste i have on it it's this one: https://www.coolblue.nl/product/767423/titan-ttg-g30030-koelpasta.html.

\

I use TG Kryonaut, which is the best when it comes to TIM before going LM.  There are better power delivery on boards, and better heat sinks yes - but we are talking about the CPU heat itself and the absorption rate of a middle of the road air cooler.

 

Same brand/make coolers do not have such a delta in performance that wouldn't be called broken, so I cant agree on that.  Its just not how it works.  You can have a low liquid AIO, but that's a defect.  A hunk of metal with a fan is still a hunk of metal with a fan.

 

In the end they don't make "golden sample" heat sinks.  I am still baffled.  Baffled enough that, as the folks on this forum know me - that Im just going to go buy a Hyper 212 Black now and debunk, refute, or readily admit Im wrong and this thing is a god of all heat sinks.  Same reason I own multiple frequency DDR4 kits - I just have to know.

 

EDIT - I just looked at the 212 Black specs, I have a similar specced (but better specs) 5 copper heat pipe air cooler Ill try first, and Ill use 4-8 different brands of fans, including SP fans.  The price for this things specs is fairly high if you ask me being its a 4 HP cooler.

[hollyh88]

10 minutes ago, Tristerin said:

I use TG Kryonaut, which is the best when it comes to TIM before going LM.  There are better power delivery on boards, and better heat sinks yes - but we are talking about the CPU heat itself and the absorption rate of a middle of the road air cooler.

 

Same brand/make coolers do not have such a delta in performance that wouldn't be called broken, so I cant agree on that.  Its just not how it works.  You can have a low liquid AIO, but that's a defect.  A hunk of metal with a fan is still a hunk of metal with a fan.

 

In the end they don't make "golden sample" heat sinks.  I am still baffled.  Baffled enough that, as the folks on this forum know me - that Im just going to go buy a Hyper 212 Black now and debunk, refute, or readily admit Im wrong and this thing is a god of all heat sinks.  Same reason I own multiple frequency DDR4 kits - I just have to know.

hey im just spitting ideas. i suppose some heatsinks even of the same brand can be manufactured just a tinnnnnyyyy bit different. some grams too much of a specific metal and you could get a better or worse version of the same heatsink. 

i mean i for one am curious if you can pull it off too. don't know how mine did it though. all i know is that my previous hyper 212 evo for the i5 7600 i had before this ryzen cpu did me well so i stayed with it. (mainly because noctua only had barf coloured coolers back then). 

i think its an amazing product for what you spend your bucks on. its known to be the budget performer. got it for 32 euros even. so pretty cheap. 

and yeah the thermal paste i have isn't the top of the line either. but i guess it performs damm well. all i know is that its thermal paste thats also used by more than just computer enthousiasts. 

 

but it wouldnt be a fair comparison if you bought the 212 black but dont have the 2600x  curious what you find though so keep me updated.