Discussion in 'AMD x86 CPUs and chipsets' started by Skramit, Nov 7, 2019.
Nice detective work!
OJR it's detective work
Following up on the Viper RAM now I've got everything running on bench.. Did some Testing with just two sticks , (both sets independently ) , and found their limits at lower clocks , i.e 3800 to be fine to around CL15-14-15-15 . around 1.41v Was able to sync IF at these clocks no problem (1900) CL14 was just a flat out no go. regardless of DRAM or SOC voltage.
Then moved to 4 Sticks (32MB ) , and found the CPU IF to now be a limitation clocking past 1833, - No booty. tried a few things, but didn't have any joy. so for now at least sticking with 1833 , 1:1 . At these clocks CL15-15-15 still 100% stable. so quite happy with that.
Might try and kick it to 1900IF at a later time and see if the CL15 timings hold up. Also have to tweak sub timings a little more.
Anyway , moral of the story, does Seem like a reasonable way to get Dual Rank 32GB with reasonable speeds and timings, @ a decent price ( $270 for the two kits with a coupon - think price is up slightly atm ) . Will be chuffed if I can
CPU Wise, Have the 5900x under an old FX 80mm AIO for now until the W/C build is done
Playing around with PBO at a max 210w PPT
Can't remember the exact PBO settings for this run.. been fiddling a lot, but
+200 or +175 PBOC
-25 CO ( not being stable on a few cores under prime and am yet to tweak each one )
-18mv or no VCORE offset
And, been playing with the XT too ....
Have you tried 3DMarks new CPU Profile test?
Id be curious to see how you went with your 5900x too.
I suspect you might get slightly better results due to your faster RAM and tighter timings (Mine its 3600/1800IF CL16-16-16-36)
I just ran a test:
Max threads 11082.0
16 threads 9691.0
8 threads 6985.0
4 threads 3731.0
2 threads 1940.0
1 thread 990.0
hey. THe beast created a new thread for it , so I posted in there
so, very similar. but without comparing clocks, hard to say (shame they don't include clock graphs in results link?
have you played around with command rate? i know when i into clocking my BDIEs command rate 1T vs 2T made a noticeable difference in benching.
im wanting to get some fast 16gb sticks, as my current ones are 2x8gb bdies, and they fly, to really see how good this 5800x is.
always more pressure to run 4 sticks then 2!
Whats teh max volts you went up too on DRAM?
i u se to have 3200 c14 8gbx8 , which i split int to 4 pairs of 2, with my 7700K Silicon Lottery sample, and with 1.88-1.91V, those sticks would do 4000 c12-12-12 1T lol....these volts not good for 24/7 use
New GSKILLS 4000 c14 kits being released which are 1.55V standard, so could maybe get your c14 3600-3733!
id also leave the IF at 1800, and leave mems at a nice round 3600mhz, and really try and tighten the tmings as far as you can with reaonable voltage
looking forward to the watercooling
more cheap Ryzen chips thanks to afterpay.
AMD Ryzen 9 5900X Desktop Processor (4.8GHz, 12 Cores, Socket AM4)(ebay.com.au) for $731.68 Delivered
AMD Ryzen 9 5950X Desktop Processor (4.9GHz, 16 Cores, Socket AM4)(ebay.com.au) for $1032.58 Delivered
AMD Ryzen 7 5800X Desktop OEM Processor (4.7GHz, 8 Cores) - TRAY CPU(ebay.com.au) for $539.75 Delivered
AMD Ryzen 7 5800X AM4 8-Core 3.80 GHz Unlocked CPU Processor(ebay.com.au) for $548.08 Delivered
AMD Ryzen 5 5600X Desktop Processor (4.6GHz, 6 Cores, Socket AM4)(ebay.com.au) for $364.48 Delivered
Finally upgraded from my 980X. Got myself a 5600X (couldn't afford the 5800X), a Gigabyte B550 Vision D-P, 32GB of Gskill Ripjaws V 3600 CL18 with a 500GB 970 evo plus for the boot drive. A nice step-up to the future.
Next upgrade is a good 10 bit ultrawide.
I see it but i dont believe its a genuine result.
CTR 2.1 RC5 thinks my 5900x is a Platinum sample.
The first run i did, it was rated as a bronze sample. Dont know what to think of this tool...
1T was very unsuccessful on these sticks for some reason. I don't think I could even run 35xx? , at least not with looser timings, would have to check my notes at home. Always a toss up in the end. Re CL14, I did think voltage scaling would help, but not up to 1.45v anyway. and this was even on 2 sticks.
realistically I should probably pump more voltage into them , just treading lightly for now
X570S boards @ mwave?
Struggling to see the point - apart from passive cooling for the chipset (which some boards already had), what else has changed?
The Old one $489 https://www.umart.com.au/AMD-AM4_C....er_attr=626126.96.36.1999188.8.131.52&mystock=1-2-7-6 so $79
windows update 30%
hahaha. very clever
You may want to lapp your 5950x for better performance
That could give some serious performance gain