F5F Stay Refreshed Hardware Desktop Gskill and Corsair DDR5 share similar timings and voltage specs, but differ in die manufacturer.

Gskill and Corsair DDR5 share similar timings and voltage specs, but differ in die manufacturer.

Gskill and Corsair DDR5 share similar timings and voltage specs, but differ in die manufacturer.

Pages (3): Previous 1 2 3 Next
D
Dominik_SK
Member
238
05-12-2021, 08:09 AM
#11
I purchased the Corsair 6000 MHz Kit today. It operates at the manufacturer's settings (6000MHz, 30-36-36-76) when using DOCP-I in BIOS. It appears to function at a 2:1 ratio instead of 1:1. The MCLK is set to 3000MHz and UCLK to 1500MHz. Shouldn't the 6000MHz model run at a 1:1 configuration by default? Or do I have to adjust it manually? Any guidance would be appreciated! Memory Timings reference: https://pasteboard.co/eoY3rgKFTzWO.png Thank you!
D
Dominik_SK
05-12-2021, 08:09 AM #11

I purchased the Corsair 6000 MHz Kit today. It operates at the manufacturer's settings (6000MHz, 30-36-36-76) when using DOCP-I in BIOS. It appears to function at a 2:1 ratio instead of 1:1. The MCLK is set to 3000MHz and UCLK to 1500MHz. Shouldn't the 6000MHz model run at a 1:1 configuration by default? Or do I have to adjust it manually? Any guidance would be appreciated! Memory Timings reference: https://pasteboard.co/eoY3rgKFTzWO.png Thank you!

O
OmegaCap777
Junior Member
1
05-12-2021, 02:58 PM
#12
Yeah, this should be in 1:1. There's probably a BIOS glitch of some sort, try putting it in 1:1 mode manually. ASUS has it somewhere in the AMD overclocking menu, I forget exactly where.
O
OmegaCap777
05-12-2021, 02:58 PM #12

Yeah, this should be in 1:1. There's probably a BIOS glitch of some sort, try putting it in 1:1 mode manually. ASUS has it somewhere in the AMD overclocking menu, I forget exactly where.

M
Maxim69rus
Member
215
05-12-2021, 08:32 PM
#13
Hello! I adjusted the UCLK DIV1 manually from AUTO to MCLK = UCLK, both set at 3000 MHz. After running an AIDA64 test, the performance looks stable. The benchmark showed speeds around 78,000 MB/s with a latency of about 80 ns. Comparing the DDR5-6000 and DDR5-6400 kits, the difference is minimal—only slight variations in throughput. It seems the 6000 kit is likely operating well, but you might want to check other parameters besides UCLK/MCLK ratio. I've attached some screenshots of the analysis for your reference.
M
Maxim69rus
05-12-2021, 08:32 PM #13

Hello! I adjusted the UCLK DIV1 manually from AUTO to MCLK = UCLK, both set at 3000 MHz. After running an AIDA64 test, the performance looks stable. The benchmark showed speeds around 78,000 MB/s with a latency of about 80 ns. Comparing the DDR5-6000 and DDR5-6400 kits, the difference is minimal—only slight variations in throughput. It seems the 6000 kit is likely operating well, but you might want to check other parameters besides UCLK/MCLK ratio. I've attached some screenshots of the analysis for your reference.

G
gr4tix
Member
70
05-12-2021, 09:17 PM
#14
The distinctions aren't clearly visible in Aida; it isn't an especially strong reference point. These effects tend to become noticeable more in games and other tasks that demand significant memory usage.
G
gr4tix
05-12-2021, 09:17 PM #14

The distinctions aren't clearly visible in Aida; it isn't an especially strong reference point. These effects tend to become noticeable more in games and other tasks that demand significant memory usage.

X
xMagicPvP
Member
154
05-14-2021, 02:46 AM
#15
The reported latency of 79.1 ns seems unusual for a standard 6000MHz kit, which typically operates in the expected range. Many users have noted variations, with some reporting values between 68 and 75 ns instead.
X
xMagicPvP
05-14-2021, 02:46 AM #15

The reported latency of 79.1 ns seems unusual for a standard 6000MHz kit, which typically operates in the expected range. Many users have noted variations, with some reporting values between 68 and 75 ns instead.

M
mr_azitium
Member
95
05-14-2021, 10:28 AM
#16
It makes sense given Aida's inconsistency, and running it again would likely reduce the score.
M
mr_azitium
05-14-2021, 10:28 AM #16

It makes sense given Aida's inconsistency, and running it again would likely reduce the score.

L
Lord_Foxtrot
Senior Member
408
05-18-2021, 01:56 AM
#17
For adjusting timing, which kits seem most straightforward for a 6000MHz run with precise settings? Would quicker-speed options (like 6600MHz) offer better binned chips versus slower ones (6000MHz)? Does a 1.35V kit provide an edge over a 1.4V one?
L
Lord_Foxtrot
05-18-2021, 01:56 AM #17

For adjusting timing, which kits seem most straightforward for a 6000MHz run with precise settings? Would quicker-speed options (like 6600MHz) offer better binned chips versus slower ones (6000MHz)? Does a 1.35V kit provide an edge over a 1.4V one?

T
Tenniskoppens
Member
183
05-18-2021, 09:43 AM
#18
In theory they could work, but in real life it's unlikely to stand out. Most Hynix units are tuned similarly, so the variation is minimal. The main distinction lies in some models handling 8000 CL34 at 1.65V while others do 8000 CL32 at the same voltage—this difference is mostly irrelevant outside competitive tests. I strongly believe G.Skill might use a lower voltage (1.35V) for their 24GB sticks because their heat spreaders aren't very effective, potentially causing more damage than benefit. This won't significantly impact performance over time if you just follow XMP settings, but manual tuning requires considering thermal effects differently than with mainstream brands like Corsair, Kingston or TeamGroup.
T
Tenniskoppens
05-18-2021, 09:43 AM #18

In theory they could work, but in real life it's unlikely to stand out. Most Hynix units are tuned similarly, so the variation is minimal. The main distinction lies in some models handling 8000 CL34 at 1.65V while others do 8000 CL32 at the same voltage—this difference is mostly irrelevant outside competitive tests. I strongly believe G.Skill might use a lower voltage (1.35V) for their 24GB sticks because their heat spreaders aren't very effective, potentially causing more damage than benefit. This won't significantly impact performance over time if you just follow XMP settings, but manual tuning requires considering thermal effects differently than with mainstream brands like Corsair, Kingston or TeamGroup.

M
mottinatorJM
Junior Member
19
05-19-2021, 06:31 PM
#19
You're experiencing stability and booting problems with your computer. It seems the RAM might be contributing, as it only supports XMP profiles and runs in DOCP I mode on your AMD system. Adjusting values manually could improve reliability. Regarding voltage settings, your BIOS allows a finer adjustment (0.00390V increments) compared to the Buildzoids guide (0.01V steps). Since your motherboard is rated at 1.4V, you should aim for 1.35V or 1.34940V. The tutorial suggests setting both CPU and DRAM voltages to 1.35V. Your kit is higher, so either option could work, but using the lower value (1.35V) might be safer. Consider setting PMIC voltages to match the recommended levels or sync all PMICs if needed. Let me know if you need further clarification.
M
mottinatorJM
05-19-2021, 06:31 PM #19

You're experiencing stability and booting problems with your computer. It seems the RAM might be contributing, as it only supports XMP profiles and runs in DOCP I mode on your AMD system. Adjusting values manually could improve reliability. Regarding voltage settings, your BIOS allows a finer adjustment (0.00390V increments) compared to the Buildzoids guide (0.01V steps). Since your motherboard is rated at 1.4V, you should aim for 1.35V or 1.34940V. The tutorial suggests setting both CPU and DRAM voltages to 1.35V. Your kit is higher, so either option could work, but using the lower value (1.35V) might be safer. Consider setting PMIC voltages to match the recommended levels or sync all PMICs if needed. Let me know if you need further clarification.

P
pillowbanana
Member
63
05-20-2021, 07:39 PM
#20
It doesn't really matter since the variation is minimal—they're essentially the same. Choose one and proceed. Target setting should be 1.4 as specified. Auto will handle synchronizing all PMICs, so either option works.
P
pillowbanana
05-20-2021, 07:39 PM #20

It doesn't really matter since the variation is minimal—they're essentially the same. Choose one and proceed. Target setting should be 1.4 as specified. Auto will handle synchronizing all PMICs, so either option works.

Pages (3): Previous 1 2 3 Next