NoddyPirate
Grand Master
I'm actually laughing at your CPU mark. All my fiddling and overclocking and undervolting got me from 21400 to 22500 on my 5600X.
Then you go and post the number for your 5800X.
ðŸ˜
Ram OC
- Thread starter Ash_
- Start date
Then you go and post the number for your 5800X.
ðŸ˜
You don’t want to know what boost clock it holds... thanks for a H150i 😜🤣I'm actually laughing at your CPU mark. All my fiddling and overclocking and undervolting got me from 21400 to 22500 on my 5600X.
Then you go and post the number for your 5800X.
ðŸ˜
NoddyPirate
Grand Master
Put me out of my misery!!
Actually, tell me your CPU temp as full load and then let me guess what the answer is!
NoddyPirate
Grand Master
OK, that's just mean.
So I'm guessing ocassional 4.95-5.0 GHz single core and sustained at 4.65-4.7 GHz all core - at something in the 60's.
How'd I do?
NoddyPirate
Grand Master
Just ran this now - not quite as good as my last attempt but I didn't save those results.....
it holds around 4.65ghz and the highest temp i have ever seen is around 65c. However, i was playing Holdfast earlier, which is very cpu heavy and it sat at 55 degrees.OK, that's just mean.
So I'm guessing ocassional 4.95-5.0 GHz single core and sustained at 4.65-4.7 GHz all core - at something in the 60's.
How'd I do?
NoddyPirate
Grand Master
It's amazing what they've done with it. The 5600X is basically a 5800X with two cores disabled and with the power dialled right down. I wish I could find a way to unlock it all!!!
So your stock all core boost is the same as my stock single core boost!
And what do you see single core? I've read some momentary boosts at low temps to 4.95 but sustained about 4.80-4.85 generally - which is still 150 MHz over advertised....
NoddyPirate
Grand Master
The other thing that amazes me is how different applications can push the CPU entirely differently. Here is what happens when the Perf Test runs it's integer and floating point tests with my Overclock in place....4.775 on all cores.....
i believe 4.95 single coreIt's amazing what they've done with it. The 5600X is basically a 5800X with two cores disabled and with the power dialled right down. I wish I could find a way to unlock it all!!!
So your stock all core boost is the same as my stock single core boost!
And what do you see single core? I've read some momentary boosts at low temps to 4.95 but sustained about 4.80-4.85 generally - which is still 150 MHz over advertised....
FerrariVie
Super Star
There you are, 3200Mhz CL22 is about what I expected:
In summary, 9.5% increase in RAM performance (according to Passmark) from 3200 CL22 to 3600 CL20 at same 1.2 stock voltage.
NoddyPirate
Grand Master
That's great. It really good to be able to validate the improvement with an Overclock. Yours looks great. Aplogies for the curveball with my query as I didn't spot that it was laptop vs desktop. Still it's very interesting to see.
Your biggest boost seems to be in read and write speeds - which shows the raw help of higher bandwidth. Interesting stuff.
NoddyPirate
Grand Master
There's a lot more than meets the eye when it comes to latency I think. The PassMark score is testing the latency from read request until the data gets all the way back to the CPU register - not just when it's available at the RAM itself - so there may be variance and smoke and mirrors going on in there somehwere.....
FerrariVie
Super Star
That is indeed weird, but it could have something to do with different architectures (Zen2 x Zen3)?
No problem at all! I also got curious to see the difference and finally had a way to measure the improvements
I tried pushing it further to 3666Mhz CL20 and while it boots, the system got really unstable with green SODs every few minutes. Maybe reducing some timings might make it become stable? But it's going to be a few weeks until I have time to play with it again. 3600 CL20 is rock solid though.
FerrariVie
Super Star
If I'm not wrong, at 3800 you loose the 1:1 clock mode to the CPU, so that why you're probably getting worse results. To keep 1:1, the farther you should go is 3733 to still get better results than with 3600.
NoddyPirate
Grand Master
Damn you @Ash_ !!!! Your scores yesterday wouldn't let me sleep last night, so I had to go fiddling more today.
Turned the Single Core Boost down a bit so I could undervolt as much as possible and managed this - 130 points up on yesterday.
All cores fully loaded with power draw at my limits and a consistent frequency at really low voltage. About as much as I can squeeze out of this CPU I think while staying at in the 60's temp wise! It's no 5800X but it's as close as I can get for now!
Turned the Single Core Boost down a bit so I could undervolt as much as possible and managed this - 130 points up on yesterday.
All cores fully loaded with power draw at my limits and a consistent frequency at really low voltage. About as much as I can squeeze out of this CPU I think while staying at in the 60's temp wise! It's no 5800X but it's as close as I can get for now!
NoddyPirate
Grand Master
I can get higher than that - but 1.35V and an extra 10 degrees for 52 MHz is a rubbish return on the investment!
I tried DLSS on cod today and wow... had a locked 220fps from my 3070Damn you @Ash_ !!!! Your scores yesterday wouldn't let me sleep last night, so I had to go fiddling more today.
Turned the Single Core Boost down a bit so I could undervolt as much as possible and managed this - 130 points up on yesterday.
View attachment 25399
All cores fully loaded with power draw at my limits and a consistent frequency at really low voltage. About as much as I can squeeze out of this CPU I think while staying at in the 60's temp wise! It's no 5800X but it's as close as I can get for now!
View attachment 25402