F5F Stay Refreshed Hardware Desktop The new OC performs well overall but struggles with synthetic tasks and doesn’t meet expectations. Why?

The new OC performs well overall but struggles with synthetic tasks and doesn’t meet expectations. Why?

The new OC performs well overall but struggles with synthetic tasks and doesn’t meet expectations. Why?

Pages (3): Previous 1 2 3 Next
T
Tico_32
Senior Member
680
03-05-2018, 02:51 AM
#11
You checked timings and bandwidth using Aida64? That tool isn't ideal here since it mainly depends on basic timing data. The updated OC performs better overall. Aida64 tends to shift results between versions, making comparisons tricky unless you stick to the same version before and after. For more accurate memory testing, consider Linpack Extreme or Y-cruncher, which handle memory well regardless of RAM size. Aida64 works fine but isn't great for real memory performance evaluation.
T
Tico_32
03-05-2018, 02:51 AM #11

You checked timings and bandwidth using Aida64? That tool isn't ideal here since it mainly depends on basic timing data. The updated OC performs better overall. Aida64 tends to shift results between versions, making comparisons tricky unless you stick to the same version before and after. For more accurate memory testing, consider Linpack Extreme or Y-cruncher, which handle memory well regardless of RAM size. Aida64 works fine but isn't great for real memory performance evaluation.

P
prestoo
Member
65
03-05-2018, 02:51 AM
#12
Intel Memory Latency Checker is available, but I’m out of the AIDA64 free trial and prefer something simpler. I ended up using benchmate instead. Please review the changes @RONOTHAN##. Right now I need to rest; I’ve been working nonstop for three days trying to resolve this issue. Not sure what’s wrong with the BIOS settings. MSI labels each VM function differently.
P
prestoo
03-05-2018, 02:51 AM #12

Intel Memory Latency Checker is available, but I’m out of the AIDA64 free trial and prefer something simpler. I ended up using benchmate instead. Please review the changes @RONOTHAN##. Right now I need to rest; I’ve been working nonstop for three days trying to resolve this issue. Not sure what’s wrong with the BIOS settings. MSI labels each VM function differently.

G
Greet_OMGtje
Junior Member
46
03-05-2018, 02:51 AM
#13
Your bandwidth rose a bit and latency also increased slightly. This might be due to the CAS latency being a bit higher at each frequency compared to your earlier 3333 configuration. Have you considered lowering the channel to CL15? It looks like you turned off gear down mode, so unless MSI altered it in their BIOS update, CAS latency should be fine.
G
Greet_OMGtje
03-05-2018, 02:51 AM #13

Your bandwidth rose a bit and latency also increased slightly. This might be due to the CAS latency being a bit higher at each frequency compared to your earlier 3333 configuration. Have you considered lowering the channel to CL15? It looks like you turned off gear down mode, so unless MSI altered it in their BIOS update, CAS latency should be fine.

X
xiPing_
Junior Member
12
03-05-2018, 02:51 AM
#14
Critical issue on TM5, but I'll attempt it again tomorrow with tCWL reduced to 12.
X
xiPing_
03-05-2018, 02:51 AM #14

Critical issue on TM5, but I'll attempt it again tomorrow with tCWL reduced to 12.

C
73
03-05-2018, 02:51 AM
#15
It seems odd that CL15 at 3533 performs better than CL14 at 3333. I didn’t realize the oddities of AMD’s odd CAS latencies.
C
CarameLJessUss
03-05-2018, 02:51 AM #15

It seems odd that CL15 at 3533 performs better than CL14 at 3333. I didn’t realize the oddities of AMD’s odd CAS latencies.

H
HootPlayz
Junior Member
8
03-05-2018, 02:51 AM
#16
So the 3533C15 isn't working yet, but 3533C14 does so far. It's only been about three minutes. I'm planning to switch to even numbers and check if it remains stable on SCL 2. I did lower it together with tRC to 53 and CWL to 14 as suggested by others.
H
HootPlayz
03-05-2018, 02:51 AM #16

So the 3533C15 isn't working yet, but 3533C14 does so far. It's only been about three minutes. I'm planning to switch to even numbers and check if it remains stable on SCL 2. I did lower it together with tRC to 53 and CWL to 14 as suggested by others.

D
DarkSkarlet
Senior Member
415
03-05-2018, 02:51 AM
#17
You need to configure 2T and turn off GDM.
D
DarkSkarlet
03-05-2018, 02:51 AM #17

You need to configure 2T and turn off GDM.

H
Hrad_Shephard
Junior Member
8
03-05-2018, 02:51 AM
#18
H
Hrad_Shephard
03-05-2018, 02:51 AM #18

I
idodi65
Member
173
03-05-2018, 02:51 AM
#19
My settings are as follows:
I
idodi65
03-05-2018, 02:51 AM #19

My settings are as follows:

D
dexterburnb
Junior Member
21
03-05-2018, 02:51 AM
#20
Im on Hynix CJR/DJR though, not Samsung B. Anyway, ive tested this new timing, and also adding linpack number by @WereCat idea. Have a look. As you can tell its not that much of a change but the performance is noticeable in synthetics and isnt run-to-run error. Y-Cruncher 1B: 50.5s PyPrime 2B: 14,8 Handbrake: 04m03s Linpack: 234,3GFlops
D
dexterburnb
03-05-2018, 02:51 AM #20

Im on Hynix CJR/DJR though, not Samsung B. Anyway, ive tested this new timing, and also adding linpack number by @WereCat idea. Have a look. As you can tell its not that much of a change but the performance is noticeable in synthetics and isnt run-to-run error. Y-Cruncher 1B: 50.5s PyPrime 2B: 14,8 Handbrake: 04m03s Linpack: 234,3GFlops

Pages (3): Previous 1 2 3 Next