• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

AMD Zen 2 (Ryzen 3000) - *** NO COMPETITOR HINTING ***

opethdisciple said:
Thanks for all the help people. The issue is I still remain at an utter loss how to get my ram stable at 3600MHz CL16.

I think the most stable I had it was when I manually entered in only the base timings (16 16 16 16 36 - 1.35v - SOC 1.1v) This remained stable for about a month but then eventually exhibited instability.

I then tried 1.4v all the way to 1.45v on the dram, with 1.45v actually making things more unstable.

But the truth is that 3200MHz CL14 is a smidgen faster than 3600MHz CL16 any way?

3200MHz CL14: 8.75ns

3600MHz CL16: 8.9ns
Click to expand...

What was the actual crash down to? Did you use event manager to see what it was? Could have been application based. I tried the new COD the other night, it froze for 30 secs and then went to desktop, but I put that down to the beta game. Same as other games or the OS, introducing patches can make the application unstable. Unless you get a sad face blue screen or the machine reboots automatically, I'd check the applications via the event manager before anything else and hardware causes are there to. After weeks of stability I'd look at software before hardware.
 
Panos said:
It doesn't factor in number of modules, sub timing etc. You need something like AIDA64 to properly measure the numbers.
No dual module RAM operated sub 70ns, since the DDR2 era.
Click to expand...

I found just the tool!

It shows if accurate that 3600MHz at CL16 is in fact better than 3200MHz at CL14.

Latency is comparable, but bandwidth is 12% higher at 3600MHz CL16.

*Although it comes with a massive disclaimer of inaccuracy. :p
 
Last edited:
Just installed new Version 5220 on my my Asus b350 plus with ABBA, does boost itself to 4425 (4.4 on the box) even with PBO disabled in the bios :eek:
1qq0wYf.png
1qq0wYf
 
I've been quite impressed with the ABBA bios getting 4.6Ghz on three cores and a significant boost to the others as well. Temps remain unchanged and voltage on the cpu is down.

Also after ABBA I was able to increase my FCLK to 1867, it wouldn't boot beyond 1800 before.
 
opethdisciple said:
Lol. :D I'm not really. I just observe using the windows task manager what speed it reports as I open up the browser and load like 20 pages.

(We should petition OcUK for more emojis!)
Click to expand...

Start HWinfo and leave it running in idle. Never heard anyone use windows to monitor anything, if windows was good then there wouldn't be a place for all these apps built and used by pros otherwise they'd all just use windows. You don't even know how often windows polls. HWinfo has been updated with AMD to get the correct polling timings. Hence why all the voltages and boosts etc were not reading right in many popular apps on release.

To be honest if I was on X370 I wouldn't be expecting the results I see on much more advanced power delivery options you see on X570 boards. Not to say you wont, but I wouldn't be expecting the same as latest tech with much more advanced VRM power delivery options if my results were lacking. I have a £160 X570 board and it takes my 3800X to 4.6+ on 5/8 cores. Playing PUBG it does 3.75-4.4 all core. Admittedly the single core boosts are when left idle with stuff running in the background but they do get there as advertised. I think people expect them to get there during testing which is another can of worms all together.
 
Woodsta888 said:
Start HWinfo and leave it running in idle. Never heard anyone use windows to monitor anything, if windows was good then there wouldn't be a place for all these apps built and used by pros otherwise they'd all just use windows. You don't even know how often windows polls. HWinfo has been updated with AMD to get the correct polling timings. Hence why all the voltages and boosts etc were not reading right in many popular apps on release.

To be honest if I was on X370 I wouldn't be expecting the results I see on much more advanced power delivery options you see on X570 boards. Not to say you wont, but I wouldn't be expecting the same as latest tech with much more advanced VRM power delivery options if my results were lacking. I have a £160 X570 board and it takes my 3800X to 4.6+ on 5/8 cores. Playing PUBG it does 3.75-4.4 all core. Admittedly the single core boosts are when left idle with stuff running in the background but they do get there as advertised. I think people expect them to get there during testing which is another can of worms all together.
Click to expand...

Don't say things like this as your make me want to buy a new mobo! I've already started looking. :cool:

But realistically, AM4 has what 1 more iteration left and we will be on a new socket, so not really worth buying a new board.

What board to you have? :D

I was looking at the Taichi x570... It would probably also sort out the issues I'm having with my ram.
 
Last edited:
opethdisciple said:
Don't say things like this as your make me want to buy a new mobo! I've already started looking. :cool:

But realistically, AM4 has what 1 more iteration left and we will be on a new socket, so not really worth buying a new board.

What board to you have? :D

I was looking at the Taichi... but if I could get a board at below £200 that essential does the same job that would be perfect. It would probably also sort out the issues I'm having with my ram.
Click to expand...

hahahah yeah, well I'd feel bad now if you still didn't get the results you were after from purchasing a new board.

Until now and particularly on intel we all played the silicon lottery. So folks would buy the CPU and an expensive board and be disappointed when the CPU was a loser, so that expensive board no matter what VRMs it had on it was wasted as the CPU was just bottom of the bin.

There are companies that but a pile of retail and test them and resell them as guaranteed clocks and add a cost to their time for testing and giving you say a 9900k that will do 5.2Ghz (at nearly £1k).

So my method gets a cheap board (latest most supported chipset) and see how it goes. Now, AMD has the guarantees on the box and I did buy a 3800X which is a binned 3700X maybe (but it does have a higher TDP available. Though mine never seems to go above 65W in HWinfo (others at 142W). If the chip performed well but not at what others were getting then if the perf uplift was worth it then I'd maybe tempted to sell that latest X570 board and get a better one.

I don't need all the extras like WiFi, on board sound and gigabit LANS which make many boards exorbitant (particularly X570) so I took a gamble and as the cheapest part between CPU & Mobo I could upgrade later if it really did noticeably under perform.

But it doesn't and it hasn't. ( Well the TDP is a concern but I get the results so I'm happy with it for now)

I'd get HWinfo and leave your PC doing nothing for a few hours and see what the max core clocks are. I'd also do some real world use like gaming and use after burner to look at the core clocks as they are in graphs to see whats happening. As so many apps and measuring tools cant seem to agree, maybe I'm bias and using the ones I like to see rather than what is real. Many apps including Ryzen master dont seem to agree so before jumping in and buying a new MOBO there maybe more scandal to come as so many apps are reading different things, but this is an architecture that works differently from what we are used to.

It's been a great upgrade from 3770k and I'm chuffed to bits with my purchases though I did go for the 8pack 3600RAM.

If it's an upgrade for you, then as long as the performance in gaming or whatever you do is there, then I wouldnt worry too much until BIOS's are mature that you are 100mhz or so off what thers will be getting. In the real world it will be a few FPS at the moment in games, nothing you'd notice expect in numbers performance wise. 3200Mhz RAm is fine and so is a 3700X.

It does play on my mind how much a MOBO BIOS team would spend getting X370 board to max that an X570. At the end of the day they'd like you to get one of their new boards.
 
I've tested my ram (8pack 3600) with tighter cl14 timings and overclocking the ram to 3800MHz, with cl16 timings. With matching 1900Mhz infinity fabric.

3600 mem lat 66.0 timespy 10,857

3800 mem lat 63.3 timespy 11,025

C20 was very slightly higher with cl14 3600.
 
You must log in or register to reply here.
Back
Top Bottom