• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.

LENOVO IdeaPad Gaming 3 15IMH05, Cinebench R20 test low results

mql

New Member
Joined
Oct 15, 2021
Messages
8 (0.15/day)
What could be the reason for such a low test result?

The test shows an average of 3070 points, with average i7-10750h processor power 63.23W
1634650118569.png


Voltage Offsets -0.070 V
1634651149375.png


To PL1 = 56W and PL2 = 92W
1634650285226.png


Core Thermal Throttling included only in PL2 Power limit
1634650423295.png


To PL1 Power limit Core ratio is only 37.0x !!!
1634650845682.png


In order for the memory modules to work stably at a CPU undervolting, the memory voltage must be increased by the same amount the CPU undervoltage.
1634653793756.png


Otherwise received IRQL_NOT_LESS_OR_EQUAL
1634651980120.png

1634653843534.png
 

Attachments

  • 10750h 1.png
    10750h 1.png
    2.9 MB · Views: 27
  • 1634651109407.png
    1634651109407.png
    29.2 KB · Views: 19
Last edited:

unclewebb

ThrottleStop & RealTemp Author
Joined
Jun 1, 2008
Messages
4,173 (0.85/day)
reason for such a low test result?
Why do you think your Cinebench result is low? The 10750H has a 45W TDP rating.


Your computer has the long term turbo power limit PL1 set to 56W. Your 3070 Cinebench R20 score is better than average for this processor. Most Dell laptops with this same processor are locked to the TDP so they cannot exceed 45W during any long term stress test.

Some computers with the 10750H have unlocked power limits and much better cooling. Some can run indefinitely at 90W. You might be able to increase your power limits some more but you are probably going to run into thermal throttling. Laptops with these CPUs are a balancing act between power limit throttling and thermal throttling.

If you have seen better Cinebench scores for the 10750H, it must have been running with unlocked power limits and better cooling.

Have you tried increasing the turbo power limits?
 

mql

New Member
Joined
Oct 15, 2021
Messages
8 (0.15/day)
Laptops with these CPUs are a balancing act between power limit throttling and thermal throttling.
I realized that this was due to the 56W limit. The increased power during a prolonged stress test causes thermal throttling. The question remains: is a 37.0x core ratio at 56W sustained power normal?
 

unclewebb

ThrottleStop & RealTemp Author
Joined
Jun 1, 2008
Messages
4,173 (0.85/day)
It is normal for a 10750H to power limit throttle and slow down during Cinebench when it is limited to 56W. Your results show that a 10750H needs somewhere around 90W so it can run at its full rated speed. This CPU will need even more than that if you are running something like Prime95 Small FFTs. There would be no easy way to keep the CPU cool when running that load.

The increased power during a prolonged stress test causes thermal throttling.
That is completely normal. The vast majority of laptops with the 9750H or 10750H either power limit throttle or thermal throttle. These are powerful mobile CPUs but very few laptop manufacturers include adequate cooling to run these CPUs indefinitely at their full rated speed.
 

mql

New Member
Joined
Oct 15, 2021
Messages
8 (0.15/day)
It is normal for a 10750H to power limit throttle and slow down during Cinebench when it is limited to 56W. Your results show that a 10750H needs somewhere around 90W so it can run at its full rated speed. This CPU will need even more than that if you are running something like Prime95 Small FFTs. There would be no easy way to keep the CPU cool when running that load.


That is completely normal. The vast majority of laptops with the 9750H or 10750H either power limit throttle or thermal throttle. These are powerful mobile CPUs but very few laptop manufacturers include adequate cooling to run these CPUs indefinitely at their full rated speed.
Thank you for the clarification.


Another question about memory stability. Using CPU undervoltage caused memory instability (IRQL_NOT_LESS_OR_EQUAL). It was necessary to unlock the memory voltage and increase the voltage as much as the processor voltage decreased. Is that how it should be? So far I have not read anywhere that such a problem exists.
 

unclewebb

ThrottleStop & RealTemp Author
Joined
Jun 1, 2008
Messages
4,173 (0.85/day)
Is that how it should be?
Reducing the CPU voltage can cause instability and BSODs. I cannot remember seeing a situation where the CPU voltage and the memory voltage were related.

An IRQL_NOT_LESS_OR_EQUAL error might often times be memory related when running the CPU at default voltage but I think it is also possible to get these errors when you start undervolting the CPU too much.

Other users with the 10750H have started losing stability at -70 mV. If you are not 100% stable at this voltage, you will need to reduce your CPU undervolt to -60 mV or -50 mV.
 

mql

New Member
Joined
Oct 15, 2021
Messages
8 (0.15/day)
Other users with the 10750H have started losing stability at -70 mV. If you are not 100% stable at this voltage, you will need to reduce your CPU undervolt to -60 mV or -50 mV.
When the memory voltage was increased by +70 mV, the stability on the CPU of -70 mV was restored.

Don't know about others, but it helped me.

I don't think I have a unique computer with a unique problem. Therefore, I suppose that in case -70mV of an unstable processor, you can try to increase the RAM voltage.
 

unclewebb

ThrottleStop & RealTemp Author
Joined
Jun 1, 2008
Messages
4,173 (0.85/day)
I think the CPU and memory voltages are completely separate. If your screenshot is correct, it shows that your memory voltage is set -80 mV lower than the recommended minimum.

1634675600449.png


The memory should be set to at least 1.350 V whether you are going to undervolt the CPU or not.
 

mql

New Member
Joined
Oct 15, 2021
Messages
8 (0.15/day)
I think the CPU and memory voltages are completely separate. If your screenshot is correct, it shows that your memory voltage is set -80 mV lower than the recommended minimum.

View attachment 221516

The memory should be set to at least 1.350 V whether you are going to undervolt the CPU or not.
For clarity. The operating voltage of the mentioned memory modules is 1.2 V. Information indicating that the mentioned operating voltage is 1.35 V has been read Aida64.

The purpose of displaying this information is to show that the RAM is currently operating at 1.270 V.
I am not aware of other programs that allow me to read the current RAM voltage. For example, CPU-Z reads the recommended profile voltage.
1634710707893.png


The intent was not to initiate a dispute about whether the RAM voltage stabilizes the processor at undervoltage. The goal was to get your opinion on the relationship between RAM voltage and CPU voltage.

Before asking about it, I checked the stability of the system several times without undervolting, only at undervoltage of the processor, and at low CPU voltage with increasing RAM voltage.
Decreasing the CPU voltage with increasing RAM voltage is most stable.
Maybe it's a feature of the motherboard, maybe it's a feature of RAM. I dont know.

I can only say that LENOVO Legion 7-15IMH05 with the same MH470 motherboard and CPU i7-10750H as LENOVO IdeaPad Gaming 3 15IMH05. The Uncore voltage offset from the factory is +70 mV.
0xBF0FE Numeric: Uncore Voltage Offset, VarStoreInfo (VarOffset/VarName): 0x1EC, VarStore: 0x2, QuestionId: 0x223, Size: 2, Min: 0x0, Max 0x3E8, Step: 0x1 {07 94 80 06 81 06 23 02 02 00 EC 01 10 11 00 00 E8 03 01 00}
0xBF112 Default: DefaultId: 0x0, Value (16 bit): 0x46 {5B 07 00 00 01 46 00}
0xBF119 End {29 02}
0xBF11B One Of: Offset Prefix, VarStoreInfo (VarOffset/VarName): 0x1EE, VarStore: 0x2, QuestionId: 0x224, Size: 1, Min: 0x0, Max 0x1, Step: 0x0 {05 91 C4 05 C5 05 24 02 02 00 EE 01 10 10 00 01 00}
0xBF12C One Of Option: +, Value (8 bit): 0x0 (default) {09 07 0C 03 30 00 00}
0xBF133 One Of Option: -, Value (8 bit): 0x1 {09 07 0D 03 00 00 01}
0xBF13A End One Of {29 02}

I know Uncore voltage does not mean the same as RAM voltage.
However, in my case, increasing the RAM voltage instead of increasing the Uncore voltage gave the best results.
 
Top