F5F Stay Refreshed Power Users Overclocking low ram write speed

low ram write speed

low ram write speed

Pages (3): Previous 1 2 3
J
JGBCRAFT
Member
65
10-20-2017, 04:39 PM
#21
Mircea Benga was surprised by the results, noting inconsistencies across different systems and questioning why writing speed remains slow on FX CPUs despite improvements.
J
JGBCRAFT
10-20-2017, 04:39 PM #21

Mircea Benga was surprised by the results, noting inconsistencies across different systems and questioning why writing speed remains slow on FX CPUs despite improvements.

R
RoseQ
Member
161
10-24-2017, 04:25 AM
#22
kerberos_20 :
Mircea Benga :
wow, you have at least double all my tests...how come?
different hardware, different result
still, it should explain you a bit why write speed is slow on FX CPUs, cpu cache is crap
Total L1 Cache
384KB
Total L2 Cache
8MB
Total L3 Cache
8MB
But there is some cache in this cpu...
R
RoseQ
10-24-2017, 04:25 AM #22

kerberos_20 :
Mircea Benga :
wow, you have at least double all my tests...how come?
different hardware, different result
still, it should explain you a bit why write speed is slow on FX CPUs, cpu cache is crap
Total L1 Cache
384KB
Total L2 Cache
8MB
Total L3 Cache
8MB
But there is some cache in this cpu...

J
Javabeanz
Member
84
10-24-2017, 07:56 AM
#23
Mircea Benga, reflecting on my previous research, your results are quite solid.
If you're concerned about power delivery, refrain from overclocking. That would boost your write speeds instead.
I have a recommendation: don’t overclock the CPU, but increase the voltage slightly within the limits your board permits. Check your write speed during different settings. This approach can reduce write times without significantly raising the voltage.
Increasing the voltage without overclocking actually enhances clock speed. However, it doesn’t change the clock frequency itself—it just affects the impact of higher voltages.
This adjustment should position you around the 8350 to 8370 range.
Kerberos 20 offers a 9590 at 5Ghz, which should yield better scores. Your 1060 seems to be running overclocked compared to its actual performance.
J
Javabeanz
10-24-2017, 07:56 AM #23

Mircea Benga, reflecting on my previous research, your results are quite solid.
If you're concerned about power delivery, refrain from overclocking. That would boost your write speeds instead.
I have a recommendation: don’t overclock the CPU, but increase the voltage slightly within the limits your board permits. Check your write speed during different settings. This approach can reduce write times without significantly raising the voltage.
Increasing the voltage without overclocking actually enhances clock speed. However, it doesn’t change the clock frequency itself—it just affects the impact of higher voltages.
This adjustment should position you around the 8350 to 8370 range.
Kerberos 20 offers a 9590 at 5Ghz, which should yield better scores. Your 1060 seems to be running overclocked compared to its actual performance.

W
W4mp3
Junior Member
10
11-01-2017, 07:27 AM
#24
I completed the test in Aida 64 with impressive speeds: Read 22344Mb/s, Write 16143 Mb/s, Copy 20443 Mb/s, Latency 68.7ns. The Maxmem2 results were concerning—11000-6000-1100, half of Aida 64's performance. I didn't realize those AMD chips had issues with their memory controllers; I wasn't even aware of what they were. I chose a random solution because all the advice was really helpful, thank you again!
W
W4mp3
11-01-2017, 07:27 AM #24

I completed the test in Aida 64 with impressive speeds: Read 22344Mb/s, Write 16143 Mb/s, Copy 20443 Mb/s, Latency 68.7ns. The Maxmem2 results were concerning—11000-6000-1100, half of Aida 64's performance. I didn't realize those AMD chips had issues with their memory controllers; I wasn't even aware of what they were. I chose a random solution because all the advice was really helpful, thank you again!

Pages (3): Previous 1 2 3