F5F Stay Refreshed Hardware Desktop Increase Micron Rev E performance from 3200 to 3600 memory speed Memory errors rise during the upgrade

Increase Micron Rev E performance from 3200 to 3600 memory speed Memory errors rise during the upgrade

Increase Micron Rev E performance from 3200 to 3600 memory speed Memory errors rise during the upgrade

Pages (2): 1 2 Next
0
0sKHD
Member
226
01-16-2019, 11:54 PM
#1
I attempted to boost my Crucial Ballistix 2x16 3200Mhz Rev E memory to 3600Mhz using XMP timings. I set the other parameters to auto values based on the kit’s specifications. The first test with Memtest86 passed, but the second failed. For Micron Rev E, I’m unsure which setting is best—whether tRCD, tRP, TRAS, tRC, or voltage matters most. If you can avoid changing voltage, focus on adjusting those timing parameters and see what stabilizes best.
0
0sKHD
01-16-2019, 11:54 PM #1

I attempted to boost my Crucial Ballistix 2x16 3200Mhz Rev E memory to 3600Mhz using XMP timings. I set the other parameters to auto values based on the kit’s specifications. The first test with Memtest86 passed, but the second failed. For Micron Rev E, I’m unsure which setting is best—whether tRCD, tRP, TRAS, tRC, or voltage matters most. If you can avoid changing voltage, focus on adjusting those timing parameters and see what stabilizes best.

P
PikaPowerA
Junior Member
29
01-24-2019, 11:16 AM
#2
First I adjusted the voltage to 1.45 to check stability. If that didn’t help, I moved on to adjusting the timing.
P
PikaPowerA
01-24-2019, 11:16 AM #2

First I adjusted the voltage to 1.45 to check stability. If that didn’t help, I moved on to adjusting the timing.

C
226
01-24-2019, 03:01 PM
#3
I recommend trying the Ryzen Dram Calculator for a solid initial estimate.
C
CrimsonGuard34
01-24-2019, 03:01 PM #3

I recommend trying the Ryzen Dram Calculator for a solid initial estimate.

H
HeyImGamerr
Member
66
01-26-2019, 09:41 AM
#4
I’d prefer not to, but I’m open to a small adjustment in timing. Maintaining longevity is important, particularly at this stage.
H
HeyImGamerr
01-26-2019, 09:41 AM #4

I’d prefer not to, but I’m open to a small adjustment in timing. Maintaining longevity is important, particularly at this stage.

M
MCkeiron
Junior Member
39
01-26-2019, 02:51 PM
#5
I tested my memory at 1.45V throughout the day. Also, Ballistix memory was running at 3600MHz with CL14. No issues, no problems.
M
MCkeiron
01-26-2019, 02:51 PM #5

I tested my memory at 1.45V throughout the day. Also, Ballistix memory was running at 3600MHz with CL14. No issues, no problems.

G
Golpaus
Junior Member
31
01-28-2019, 03:06 AM
#6
The Ballistix ICs remain unaffected by temperature, operating reliably up to around 70°C. Most users prefer the 1.5V configuration daily, though safety drops near 1.7V. Only frequency and tCL are directly influenced by voltage changes—raising power improves those two parameters more noticeably. The DRAM Calculator often misrepresents things; instead of focusing on CL14 tuning, consider switching to CL15 with GDM disabled. Confirm your setup by loading Thaiphoon Thruster to inspect the PCB revision or check underneath the heat spreaders for multiple DRAM wafers. Clocking performance isn’t the only concern—timings should be adjusted carefully. A standard Ballistix could easily exceed 3600 MHz, but this depends heavily on your IMCs. The main issue likely stems from suboptimal auto-timing settings. Start by loosening all clock settings significantly, especially primary and secondary timings. Modern systems prioritize bandwidth, so higher FCLK usually helps latency. Aim for a stable 1867–2000 MHz FCLK with MCLK and UCLK aligned. Adjust BIOS to set a manual FCLK of 1866, VDIMM/DRAM voltage at 1.5V, and SoC voltage at 1.15V. Use Medium LLC if needed, and consider a 1.25V supply for better stability. Avoid memtest86 due to slow error detection; prefer tools like Anta-Profile on TM5 or OCCT with SSE/AVX2 for thorough testing. Run a 30-minute SSE OCCT and 30 minutes of AVX2 testing to verify RAM, CPU, and timing health. If stable at 2000 MHz, push to 1933 or higher. For B550, newer revisions (like B550+ Dark Hero) seem more consistent with 2000 MHz. Once stable, proceed to fine-tune timings—start with tCL around 20, tRCDRD 22, and gradually tighten. If you're dual-rank, leave tRFC loose initially and focus on primary clock alignment.
G
Golpaus
01-28-2019, 03:06 AM #6

The Ballistix ICs remain unaffected by temperature, operating reliably up to around 70°C. Most users prefer the 1.5V configuration daily, though safety drops near 1.7V. Only frequency and tCL are directly influenced by voltage changes—raising power improves those two parameters more noticeably. The DRAM Calculator often misrepresents things; instead of focusing on CL14 tuning, consider switching to CL15 with GDM disabled. Confirm your setup by loading Thaiphoon Thruster to inspect the PCB revision or check underneath the heat spreaders for multiple DRAM wafers. Clocking performance isn’t the only concern—timings should be adjusted carefully. A standard Ballistix could easily exceed 3600 MHz, but this depends heavily on your IMCs. The main issue likely stems from suboptimal auto-timing settings. Start by loosening all clock settings significantly, especially primary and secondary timings. Modern systems prioritize bandwidth, so higher FCLK usually helps latency. Aim for a stable 1867–2000 MHz FCLK with MCLK and UCLK aligned. Adjust BIOS to set a manual FCLK of 1866, VDIMM/DRAM voltage at 1.5V, and SoC voltage at 1.15V. Use Medium LLC if needed, and consider a 1.25V supply for better stability. Avoid memtest86 due to slow error detection; prefer tools like Anta-Profile on TM5 or OCCT with SSE/AVX2 for thorough testing. Run a 30-minute SSE OCCT and 30 minutes of AVX2 testing to verify RAM, CPU, and timing health. If stable at 2000 MHz, push to 1933 or higher. For B550, newer revisions (like B550+ Dark Hero) seem more consistent with 2000 MHz. Once stable, proceed to fine-tune timings—start with tCL around 20, tRCDRD 22, and gradually tighten. If you're dual-rank, leave tRFC loose initially and focus on primary clock alignment.

I
InsygniaS
Member
55
01-28-2019, 05:01 AM
#7
Great progress! You successfully achieved stable results using default voltages, but OCCT continues to report issues. You've adjusted the DRAM voltage to 1.4 and are re-running OCCT Memtest.
I
InsygniaS
01-28-2019, 05:01 AM #7

Great progress! You successfully achieved stable results using default voltages, but OCCT continues to report issues. You've adjusted the DRAM voltage to 1.4 and are re-running OCCT Memtest.

1
19ten
Junior Member
40
01-28-2019, 09:27 AM
#8
Also verified Taiphoon Burner is indeed Rev E or E-Die. I’ve already seen it’s Dual Rank, and the product info ends with M16FE which confirms Rev E. Thanks for all the details—will keep an eye on it but aim to hit 3600 soon without a big timing hit.
1
19ten
01-28-2019, 09:27 AM #8

Also verified Taiphoon Burner is indeed Rev E or E-Die. I’ve already seen it’s Dual Rank, and the product info ends with M16FE which confirms Rev E. Thanks for all the details—will keep an eye on it but aim to hit 3600 soon without a big timing hit.

P
pedro_tkf
Senior Member
643
01-29-2019, 12:18 PM
#9
Alright, so far no errors after half an hour of OCCT memory test with CL 16, tRCDRD 19, tRCDWR 18, tRP 18, tRAS 38 and tRC 58 @1.4v . Everything else is still on auto. I'll let it run the full hour, though, just to make sure.
P
pedro_tkf
01-29-2019, 12:18 PM #9

Alright, so far no errors after half an hour of OCCT memory test with CL 16, tRCDRD 19, tRCDWR 18, tRP 18, tRAS 38 and tRC 58 @1.4v . Everything else is still on auto. I'll let it run the full hour, though, just to make sure.

_
_Ace_Plays
Member
55
02-11-2019, 09:43 PM
#10
No, I'm not at that speed. My capabilities are designed for a different range.
_
_Ace_Plays
02-11-2019, 09:43 PM #10

No, I'm not at that speed. My capabilities are designed for a different range.

Pages (2): 1 2 Next