F5F Stay Refreshed Hardware Desktop DDR4 Gear 1 instability issue?

DDR4 Gear 1 instability issue?

DDR4 Gear 1 instability issue?

A
Amtrak10
Senior Member
639
03-26-2016, 10:03 PM
#1
Environmental conditions and component health influence stability. Temperature, power supply consistency, and fan performance play key roles. Hardware wear or voltage fluctuations may cause crashes under stress.
A
Amtrak10
03-26-2016, 10:03 PM #1

Environmental conditions and component health influence stability. Temperature, power supply consistency, and fan performance play key roles. Hardware wear or voltage fluctuations may cause crashes under stress.

A
Avogadroz
Junior Member
7
03-28-2016, 08:48 PM
#2
Several voltage levels influence its performance, with VCCSA being the primary factor. Increasing it to 1.2V might stabilize it. Other relevant voltages include VCCIO2, though VCCSA is the main one to adjust. The DRAM voltage could also be too low; DDR4 uses it for both memory and controller functions, potentially requiring over 1.3V for 3200 to operate properly. Testing at 1.35V may help improve the situation.
A
Avogadroz
03-28-2016, 08:48 PM #2

Several voltage levels influence its performance, with VCCSA being the primary factor. Increasing it to 1.2V might stabilize it. Other relevant voltages include VCCIO2, though VCCSA is the main one to adjust. The DRAM voltage could also be too low; DDR4 uses it for both memory and controller functions, potentially requiring over 1.3V for 3200 to operate properly. Testing at 1.35V may help improve the situation.

G
GamerBlu
Member
63
04-07-2016, 09:24 PM
#3
Gear 1 stabilized at a DRAM voltage of 1.35V. Raised VCCSA and VCCIO2 to 1.2V and optimized timings. Your thoughts on these settings? Please note this is mismatched OEM memory, not a high-end matched kit. Would you consider pushing for higher frequencies? The official max speed for Gear 1 on this CPU is 3200, while other SKUs from the same generation are around 2933 at that setting. Do you have any experience with Gear 1 running at 3200 with different models? Current memory testing tools include Memtest86+ and others. I used it a long time ago.
G
GamerBlu
04-07-2016, 09:24 PM #3

Gear 1 stabilized at a DRAM voltage of 1.35V. Raised VCCSA and VCCIO2 to 1.2V and optimized timings. Your thoughts on these settings? Please note this is mismatched OEM memory, not a high-end matched kit. Would you consider pushing for higher frequencies? The official max speed for Gear 1 on this CPU is 3200, while other SKUs from the same generation are around 2933 at that setting. Do you have any experience with Gear 1 running at 3200 with different models? Current memory testing tools include Memtest86+ and others. I used it a long time ago.

N
NikitaVTF2480
Junior Member
12
04-07-2016, 10:49 PM
#4
You can usually tighten tRCD slightly, but it often caps around 20 or goes up to at least 18 or 19. tRAS tends to be quite loose, with few cases where it doesn’t hit 28—unless you’re using Micron memory. In practice, it’s not the most critical timing issue because it’s rarely used that way, making it probably not worth the effort to fine-tune. If you gather opinions from others, you’ll likely get varied results. My preferred tools include TestMem5, Y-Cruncher’s test suite (especially the VST test), and HCI Memtest. Some people also recommend MemTest86, Kahru, MemtestPro, OCCT, or similar. When overclocking memory, I usually start with VST for an hour to discover the max frequency. If that fails, I run Testmem5 to adjust timings (Anta777 Extreme1 preset for DDR4, 1usmus_v3 for DDR5, at least two hours or more), then continue with HCI Memtest for extended periods—often around 1000%, though for DDR5 I’ve gone higher. Asking others will give you different advice, which is just what I’ve found effective for achieving full stability.
N
NikitaVTF2480
04-07-2016, 10:49 PM #4

You can usually tighten tRCD slightly, but it often caps around 20 or goes up to at least 18 or 19. tRAS tends to be quite loose, with few cases where it doesn’t hit 28—unless you’re using Micron memory. In practice, it’s not the most critical timing issue because it’s rarely used that way, making it probably not worth the effort to fine-tune. If you gather opinions from others, you’ll likely get varied results. My preferred tools include TestMem5, Y-Cruncher’s test suite (especially the VST test), and HCI Memtest. Some people also recommend MemTest86, Kahru, MemtestPro, OCCT, or similar. When overclocking memory, I usually start with VST for an hour to discover the max frequency. If that fails, I run Testmem5 to adjust timings (Anta777 Extreme1 preset for DDR4, 1usmus_v3 for DDR5, at least two hours or more), then continue with HCI Memtest for extended periods—often around 1000%, though for DDR5 I’ve gone higher. Asking others will give you different advice, which is just what I’ve found effective for achieving full stability.

M
Maail99
Member
161
04-08-2016, 07:47 AM
#5
tRCD isn't adjusting anymore, despite reducing the setting—it won't drop below 19. Yes, two of the four rods are Micron, and it doesn't stabilize at tRAS when lowered further.
M
Maail99
04-08-2016, 07:47 AM #5

tRCD isn't adjusting anymore, despite reducing the setting—it won't drop below 19. Yes, two of the four rods are Micron, and it doesn't stabilize at tRAS when lowered further.