F5F Stay Refreshed Hardware Desktop The genuine advantages of a high SP rating lie in its reliability and performance consistency.

The genuine advantages of a high SP rating lie in its reliability and performance consistency.

The genuine advantages of a high SP rating lie in its reliability and performance consistency.

N
NottaSpy
Member
232
01-08-2021, 08:01 PM
#1
Through some investigation, I understand the SP score serves as an automated measure of silicon quality based only on the voltage table. It’s generally reliable, though it isn’t the definitive conclusion. I’ve noticed others pointing out that chips with lower scores sometimes performed better overall. For example, during der8auer’s 12900KS review, a SP80 chip ran at 5.2GHz at 1.274V, whereas an SP88 chip achieved the same speed at 1.282V. My view is that SP mainly reflects voltage efficiency rather than processor scaling, so it shouldn’t be relied upon as a performance gauge. Of course, voltage differences between cores are minor and don’t significantly impact overclock limits. Anyone else agree with this perspective or do you feel off?
N
NottaSpy
01-08-2021, 08:01 PM #1

Through some investigation, I understand the SP score serves as an automated measure of silicon quality based only on the voltage table. It’s generally reliable, though it isn’t the definitive conclusion. I’ve noticed others pointing out that chips with lower scores sometimes performed better overall. For example, during der8auer’s 12900KS review, a SP80 chip ran at 5.2GHz at 1.274V, whereas an SP88 chip achieved the same speed at 1.282V. My view is that SP mainly reflects voltage efficiency rather than processor scaling, so it shouldn’t be relied upon as a performance gauge. Of course, voltage differences between cores are minor and don’t significantly impact overclock limits. Anyone else agree with this perspective or do you feel off?

D
deathtodawn
Member
216
01-12-2021, 02:55 PM
#2
These silicon scores, such as SP or the outdated "ASIC quality" reported by GPU-Z for certain GPUs, are merely rough estimates at best and often completely irrelevant at worst. The true measure of silicon performance comes only from thorough testing—pushing it to its limits in various ways to uncover its true capabilities and limitations. For instance, my old RX 580 with 4GB RAM struggles to reach even 2000 MHz, and the core rarely hits 1400 MHz under normal settings. While it does undervolt to -60mV while keeping the stock speed of 1386 MHz, that’s still within the realm of what’s technically possible as an OC+UV. Ultimately, no single number can accurately describe such a configuration; you must experiment until you discover what works best for your needs.
D
deathtodawn
01-12-2021, 02:55 PM #2

These silicon scores, such as SP or the outdated "ASIC quality" reported by GPU-Z for certain GPUs, are merely rough estimates at best and often completely irrelevant at worst. The true measure of silicon performance comes only from thorough testing—pushing it to its limits in various ways to uncover its true capabilities and limitations. For instance, my old RX 580 with 4GB RAM struggles to reach even 2000 MHz, and the core rarely hits 1400 MHz under normal settings. While it does undervolt to -60mV while keeping the stock speed of 1386 MHz, that’s still within the realm of what’s technically possible as an OC+UV. Ultimately, no single number can accurately describe such a configuration; you must experiment until you discover what works best for your needs.

A
Alas5
Junior Member
3
01-12-2021, 08:23 PM
#3
The SP rating for later-gen Intel processors is a valid concept.
A
Alas5
01-12-2021, 08:23 PM #3

The SP rating for later-gen Intel processors is a valid concept.

A
AliReyiz_IK
Member
113
01-12-2021, 09:18 PM
#4
I was just referencing "ASIC quality" as another case where a number seems outdated and misleading—people gave it more importance than it deserved, so it no longer appears in reports. I don’t believe SP really holds much significance. It might still have some minor relevance, but it can’t substitute direct testing of overclocks or undervolts.
A
AliReyiz_IK
01-12-2021, 09:18 PM #4

I was just referencing "ASIC quality" as another case where a number seems outdated and misleading—people gave it more importance than it deserved, so it no longer appears in reports. I don’t believe SP really holds much significance. It might still have some minor relevance, but it can’t substitute direct testing of overclocks or undervolts.

X
xJuanSolo
Junior Member
44
01-12-2021, 10:03 PM
#5
The ASIC performance on an older GTX 480 was around 47%, but it still managed a steady 960MHz core clock in air. The ASIC concept didn’t really matter.
X
xJuanSolo
01-12-2021, 10:03 PM #5

The ASIC performance on an older GTX 480 was around 47%, but it still managed a steady 960MHz core clock in air. The ASIC concept didn’t really matter.