F5F Stay Refreshed Power Users Overclocking People apply an offset for AVX during OC to align the data correctly.

People apply an offset for AVX during OC to align the data correctly.

People apply an offset for AVX during OC to align the data correctly.

Pages (3): Previous 1 2 3 Next
X
XxGrenidierXx
Posting Freak
813
05-30-2023, 07:34 AM
#11
It shouldn't be thermal throttling at those temperatures. Which encoder were you employing in HandBrake? For example, x264, x265 (also known as HEVC), etc.
X
XxGrenidierXx
05-30-2023, 07:34 AM #11

It shouldn't be thermal throttling at those temperatures. Which encoder were you employing in HandBrake? For example, x264, x265 (also known as HEVC), etc.

A
angelcake_11
Senior Member
540
05-30-2023, 01:05 PM
#12
i am not fully versed in how handbrake uses the cpu overall. i do know that ram, hdd speed and other factors influence how it works. it's likely that at the higher speed your hitting a limiting factor from elsewhere so the program has to slow a touch to allow other parts of the pc to keep up.
back in the day when i used to do a lot of video work and such i always kept the temp files i was working with on a separate physical drive than where the finished file went. remember the hdd can only write or read at one time. not both. so compressing to the same drive means it has to read data and then write data in alternating fashion. going from one to another drive allows it to read fully from one and then write fully to the other. if i recall right i saw a good 50% increase in frames/second processed when i kept 2 drives working. this keeps the ram from filling up and having to wait for the hdd so it can offload the finished data dn then load new raw data and so on.
it's the old "weakest link" type scenario.
A
angelcake_11
05-30-2023, 01:05 PM #12

i am not fully versed in how handbrake uses the cpu overall. i do know that ram, hdd speed and other factors influence how it works. it's likely that at the higher speed your hitting a limiting factor from elsewhere so the program has to slow a touch to allow other parts of the pc to keep up.
back in the day when i used to do a lot of video work and such i always kept the temp files i was working with on a separate physical drive than where the finished file went. remember the hdd can only write or read at one time. not both. so compressing to the same drive means it has to read data and then write data in alternating fashion. going from one to another drive allows it to read fully from one and then write fully to the other. if i recall right i saw a good 50% increase in frames/second processed when i kept 2 drives working. this keeps the ram from filling up and having to wait for the hdd so it can offload the finished data dn then load new raw data and so on.
it's the old "weakest link" type scenario.

M
MarderRyer
Junior Member
16
06-03-2023, 12:50 AM
#13
Ralph,
stress evaluations differ greatly and can be split into two types; stability tests involving changing workloads, and thermal tests with constant loads. Intel runs their processors under a consistent 100% TDP load to confirm Thermal Specifications. Prime95 version 26.6 Small FFT’s is perfect for CPU thermal analysis since it maintains a steady 100% workload, keeping core temps within +/- a few percent of TDP. It closely mimics Intel’s conditions without exceeding them. Other tools that don’t push or limit the processor will provide a reliable thermal baseline. A comparison of utilities sorted by thermal and stability tests based on % of TDP, averaged across six processor generations at default settings (rounded to five%), reveals significant variations in workloads between Prime95 AVX versions like 29.4 versus non-AVX 26.6. All evaluations will display 100% CPU utilization in Windows Task Manager, reflecting processor usage rather than % TDP. Core temperatures align with power consumption (Watts), which equals workload. Prime95 v26.6 Small FFT’s delivers a consistent 100% load, meaning if core temps stay below 85°C, the chip should handle demanding real-world tasks without overheating.

Prime95 versions after 26.6 utilize AVX/2 instructions in the CPU’s Floating Point Unit (FPU), creating an impractical workload with core temps up to 20°C higher than Prime95 v26.6. To keep temperatures at 85°C or lower, an AVX offset of -3 (300 MHz) or greater might be necessary. Even without AVX applications, the BIOS must support it, as some utilities rely on AVX for stability checks.

AVX can be turned off in Prime95 versions beyond 26.6 by adding "CpuSupportsAVX=0" to the local.txt file in the Prime95 directory. But since core temperatures remain similar to 26.6, sticking with version 26.6 is simpler.

According to Intel’s datasheets, TDP and thermal specs are verified “without AVX.” Refer to page 87, Thermal Considerations, Section 5.1.1, first paragraph, second sentence.

8th and 9th Generation Datasheet, Volume 1 -
https://www.intel.com/content/www/us/en/...vol-1.html
M
MarderRyer
06-03-2023, 12:50 AM #13

Ralph,
stress evaluations differ greatly and can be split into two types; stability tests involving changing workloads, and thermal tests with constant loads. Intel runs their processors under a consistent 100% TDP load to confirm Thermal Specifications. Prime95 version 26.6 Small FFT’s is perfect for CPU thermal analysis since it maintains a steady 100% workload, keeping core temps within +/- a few percent of TDP. It closely mimics Intel’s conditions without exceeding them. Other tools that don’t push or limit the processor will provide a reliable thermal baseline. A comparison of utilities sorted by thermal and stability tests based on % of TDP, averaged across six processor generations at default settings (rounded to five%), reveals significant variations in workloads between Prime95 AVX versions like 29.4 versus non-AVX 26.6. All evaluations will display 100% CPU utilization in Windows Task Manager, reflecting processor usage rather than % TDP. Core temperatures align with power consumption (Watts), which equals workload. Prime95 v26.6 Small FFT’s delivers a consistent 100% load, meaning if core temps stay below 85°C, the chip should handle demanding real-world tasks without overheating.

Prime95 versions after 26.6 utilize AVX/2 instructions in the CPU’s Floating Point Unit (FPU), creating an impractical workload with core temps up to 20°C higher than Prime95 v26.6. To keep temperatures at 85°C or lower, an AVX offset of -3 (300 MHz) or greater might be necessary. Even without AVX applications, the BIOS must support it, as some utilities rely on AVX for stability checks.

AVX can be turned off in Prime95 versions beyond 26.6 by adding "CpuSupportsAVX=0" to the local.txt file in the Prime95 directory. But since core temperatures remain similar to 26.6, sticking with version 26.6 is simpler.

According to Intel’s datasheets, TDP and thermal specs are verified “without AVX.” Refer to page 87, Thermal Considerations, Section 5.1.1, first paragraph, second sentence.

8th and 9th Generation Datasheet, Volume 1 -
https://www.intel.com/content/www/us/en/...vol-1.html

S
SpectatorSEBI_
Junior Member
22
06-03-2023, 09:14 AM
#14
appreciate the insight, it gives me a lot of information, but I still don't understand (and likely won't until I get home this evening to test). Without AVX offset, what could have restricted the CPU to 83-87% load? HandBrake, which uses AVX, is a major consumer and keeps all cores at least 95-100%, usually higher. Regardless of whether I'm working on a low resolution or low FPS cartoon, it will still reach 97-100%. Based on what I've heard from either Exroofer or Math Geek, could the current throttling be the reason for the load? Also, my Vcore is set to 1.240V and Fixed.
S
SpectatorSEBI_
06-03-2023, 09:14 AM #14

appreciate the insight, it gives me a lot of information, but I still don't understand (and likely won't until I get home this evening to test). Without AVX offset, what could have restricted the CPU to 83-87% load? HandBrake, which uses AVX, is a major consumer and keeps all cores at least 95-100%, usually higher. Regardless of whether I'm working on a low resolution or low FPS cartoon, it will still reach 97-100%. Based on what I've heard from either Exroofer or Math Geek, could the current throttling be the reason for the load? Also, my Vcore is set to 1.240V and Fixed.

N
Nessa106
Member
169
06-03-2023, 10:09 AM
#15
It shouldn't be thermal throttling at those temperatures. Which encoder were you employing in HandBrake? For example, x264, x265 (also known as HEVC), etc.
N
Nessa106
06-03-2023, 10:09 AM #15

It shouldn't be thermal throttling at those temperatures. Which encoder were you employing in HandBrake? For example, x264, x265 (also known as HEVC), etc.

K
khaledkb_
Senior Member
724
06-03-2023, 04:33 PM
#16
I don't think x264 utilizes AVX, which might explain the lower power usage and temperatures even without an AVX offset. You might want to explore x265 instead if you're curious about using AVX.
K
khaledkb_
06-03-2023, 04:33 PM #16

I don't think x264 utilizes AVX, which might explain the lower power usage and temperatures even without an AVX offset. You might want to explore x265 instead if you're curious about using AVX.

E
Eppikx
Senior Member
447
06-04-2023, 12:58 AM
#17
The Original Ralph :
... without AVX offset, the CPU would have been restricted to 83-87% utilization. HandBrake (which uses AVX) acts as a central power consumer and manages all cores ... 97-100% ...
It's important to recognize that "CPU utilization" merely reflects processor activity, not the actual TDP workload. To grasp % TDP accurately, consider power consumption in watts. Based on Intel's default configurations, your 9900K should consume roughly 95 Watts plus or minus a few percent as specified, when tested with non-AVX Prime95 Small FFT. But factors like VID, leakage current, ambient temperature, and any cooling methods outside the stock range can cause TDP to differ slightly.
When employing custom BIOS settings such as MCE, manual or adaptive Vcore, LLC or other overclock options, under identical workloads (non-AVX Prime95 Small FFT), power usage surpasses the rated TDP, while CPU utilization stays below 100%. Therefore, to better grasp what your CPU is experiencing during different tasks, concentrate on power output instead of utilization. This approach helps identify the exact power thresholds where issues arise, greatly supporting troubleshooting and performance analysis.
E
Eppikx
06-04-2023, 12:58 AM #17

The Original Ralph :
... without AVX offset, the CPU would have been restricted to 83-87% utilization. HandBrake (which uses AVX) acts as a central power consumer and manages all cores ... 97-100% ...
It's important to recognize that "CPU utilization" merely reflects processor activity, not the actual TDP workload. To grasp % TDP accurately, consider power consumption in watts. Based on Intel's default configurations, your 9900K should consume roughly 95 Watts plus or minus a few percent as specified, when tested with non-AVX Prime95 Small FFT. But factors like VID, leakage current, ambient temperature, and any cooling methods outside the stock range can cause TDP to differ slightly.
When employing custom BIOS settings such as MCE, manual or adaptive Vcore, LLC or other overclock options, under identical workloads (non-AVX Prime95 Small FFT), power usage surpasses the rated TDP, while CPU utilization stays below 100%. Therefore, to better grasp what your CPU is experiencing during different tasks, concentrate on power output instead of utilization. This approach helps identify the exact power thresholds where issues arise, greatly supporting troubleshooting and performance analysis.

M
230
06-06-2023, 03:43 AM
#18
Thanks - that's what I've been using.
M
MaddieStarr801
06-06-2023, 03:43 AM #18

Thanks - that's what I've been using.

P
Puppy_Power4
Member
167
06-11-2023, 04:46 AM
#19
I don't think x264 uses AVX, which might explain the lower power usage and temperatures even without an AVX offset. I think x265 likely does use AVX, since the clock ratio was 48 or 4.8 OC. I assumed h264 probably includes AVX because the clock was set to 4.8, and I adjusted the avx offset to -2. Most cores were running around 4622 (4.6), occasionally jumping to 4.8 before dropping back to 4.6. So it seems they could handle tasks without using those instructions.
P
Puppy_Power4
06-11-2023, 04:46 AM #19

I don't think x264 uses AVX, which might explain the lower power usage and temperatures even without an AVX offset. I think x265 likely does use AVX, since the clock ratio was 48 or 4.8 OC. I assumed h264 probably includes AVX because the clock was set to 4.8, and I adjusted the avx offset to -2. Most cores were running around 4622 (4.6), occasionally jumping to 4.8 before dropping back to 4.6. So it seems they could handle tasks without using those instructions.

L
Lake902
Junior Member
4
06-11-2023, 05:19 AM
#20
Ah, ok, learn something new every day!
L
Lake902
06-11-2023, 05:19 AM #20

Ah, ok, learn something new every day!

Pages (3): Previous 1 2 3 Next