Jump to content

Intel allows memory overclocking on B660 and H670, but only sort of.

TLDR: If you want to run 3200MHz+ DDR4 on Alder Lake, there's a chance you can't run it if you don't buy the unlocked (K sku) variant regardless of motherboard support for memory overclocking. If it's 4000MHz+, you can give up. If you already placed your orders or having issues of this already, try make the CPU run cooler.

 

Source 1: https://www.chiphell.com/thread-2385317-1-1.html

 

Source 2: https://www.bilibili.com/video/BV18L4y1t7DW?share_source=copy_web

 

As everyone knows, Intel opened up memory overclocking to some non-Z/X chipsets since their 500 series chipset with the B560 and H570. It's not on par with AMD yet since you still can't overclock CPUs (let alone buying K skus also being a requirement), but it still brings cheaper motherboards to those who just want to use faster memory.

 

But it seems like Intel is quite bitter about this and made their move on Alder Lake (12th gen).

 

Quote and translated source 1 (simplified to keep things brief)

Quote

> The system used is the i7-12700 and MSI B660M Mortar with Windows 11 powered by a Seasonic Prime 750, no graphics card.

 

> The system crashed (screen went blank) during CPU-Z benchmark (T/N: a relatively light benchmark) after freshly installing the OS

 

> The user thought it's because he only used 1 CPU 8pin instead of 2, so he connected the remaining connector but it crashed even more including blue screens.

 

> Determined it's a memory error from fault codes returned by the OS

 

> The memory involved in a set of dual rank per stick Crucial Ballistix 3600MHz CL16 2x16GB kit at default settings. The same memory kit achieved 4000MHz CL16 with an AMD R7 5700G so he doesn't think the memory kit is this bad.

 

> The user can't think of a different solution, so he swapped to a single rank per stick Crucial 3600MHz CL16 2x16GB kit. Also using default XMP settings, AIDA64 memory test failed after 30 mins

 

> Looking at the applied settings, he determined that the default CPU SA (system agent) voltage is too low at just 0.92V. Alder Lake OC guides (T/N: which one is not mentioned) typically recommend 1.2V if not more but he found out the MSI board doesn't let him raise it at all.

 

> The user bought an Asus TUF B660M-Plus Wifi D4 motherboard and tried again, yet SA voltage still wouldn't buldge

 

Now it doesn't seem to be an universal issue for CPUs to crash at just 3600MHz memory with stock SA voltage. However because Alder Lake CPUs support up to 3200MHz DDR4 by Intel spec (i.e. 3200MHz DDR4 is definitely fine for the CPU), it does mean there's very little memory overclocking headroom left if you lose the silicon lottery. Even if you win, non K CPUs will still be far behind the K models because K skus on B660 and H670 motherboards have unlocked SA voltage but non K does not. At this point I should add that such limitation didn't exist on B560 and H570, you could pair any CPU with them and SA voltage is free for you to adjust unless the board maker locked you out (which could happen on prebuilts with custom motherboard, but not motherboards sold independently at retail)

 

Not going to translate source 2 because I'm tired, but all that I want to quote from it is that lowering the CPU temperatures do seem to make the CPU more capable in running faster memory. In the video the author had a crashing system with the stock cooler. By maxing out the fans, it managed to stay stable. 

 

Now you might be wondering, can this change in the future? Well unfortunately, only Intel can do something about it because board makers cannot. The SA voltage is regulated by a internal regulator inside the CPU that also controls everything else except the CPU core voltage. Board makers only have control over voltage generated on the motherboard, not voltage regulated inside the CPU. At the same time because overclocking pretty much mandates a K sku when non K does this badly, Intel now grabs more cash to themselves than ever before. In the past before B and some H chipsets can overclock memory, it's not uncommon (especially when that memory standard gets older) to pair a Z chipset motherboard with a non K CPU for only memory overclocking capability. Of course users spend more for this, but just the motherboard which means Intel hardly earned any more money from it. Now, Intel gets the cash while board makers... don't. What we have now then is "only Intel can change this" and "only Intel benefits from this", so not only there's no alternative way out, the gate keeper is the one who would reject this idea the most. Therefore, this is near impossible to change. Of course I welcome Intel to change their mind.

CPU: i7-2600K 4751MHz 1.44V (software) --> 1.47V at the back of the socket Motherboard: Asrock Z77 Extreme4 (BCLK: 103.3MHz) CPU Cooler: Noctua NH-D15 RAM: Adata XPG 2x8GB DDR3 (XMP: 2133MHz 10-11-11-30 CR2, custom: 2203MHz 10-11-10-26 CR1 tRFC:230 tREFI:14000) GPU: Asus GTX 1070 Dual (Super Jetstream vbios, +70(2025-2088MHz)/+400(8.8Gbps)) SSD: Samsung 840 Pro 256GB (main boot drive), Transcend SSD370 128GB PSU: Seasonic X-660 80+ Gold Case: Antec P110 Silent, 5 intakes 1 exhaust Monitor: AOC G2460PF 1080p 144Hz (150Hz max w/ DP, 121Hz max w/ HDMI) TN panel Keyboard: Logitech G610 Orion (Cherry MX Blue) with SteelSeries Apex M260 keycaps Mouse: BenQ Zowie FK1

 

Model: HP Omen 17 17-an110ca CPU: i7-8750H (0.125V core & cache, 50mV SA undervolt) GPU: GTX 1060 6GB Mobile (+80/+450, 1650MHz~1750MHz 0.78V~0.85V) RAM: 8+8GB DDR4-2400 18-17-17-39 2T Storage: HP EX920 1TB PCIe x4 M.2 SSD + Crucial MX500 1TB 2.5" SATA SSD, 128GB Toshiba PCIe x2 M.2 SSD (KBG30ZMV128G) gone cooking externally, 1TB Seagate 7200RPM 2.5" HDD (ST1000LM049-2GH172) left outside Monitor: 1080p 126Hz IPS G-sync

 

Desktop benching:

Cinebench R15 Single thread:168 Multi-thread: 833 

SuperPi (v1.5 from Techpowerup, PI value output) 16K: 0.100s 1M: 8.255s 32M: 7m 45.93s

Link to comment
Share on other sites

Link to post
Share on other sites

Wouldnt a voltmod fix this? Not software mod but hardmod the board with a potentiometer to raise and decrease sa volt

 

or is sa volt derived from a voltage reference and only the cpu can internally change it?

Link to comment
Share on other sites

Link to post
Share on other sites

2 minutes ago, Somerandomtechyboi said:

Wouldnt a voltmod fix this? Not software mod but hardmod the board with a potentiometer to raise and decrease sa volt

 

or is sa volt derived from a voltage reference and only the cpu can internally change it?

It's the FIVR approach like on Skylake-X. Whatever you do on the outside will change things you want to keep unchanged.

 

Plus compared to a volt mod, it will be more practical to buy a K sku

CPU: i7-2600K 4751MHz 1.44V (software) --> 1.47V at the back of the socket Motherboard: Asrock Z77 Extreme4 (BCLK: 103.3MHz) CPU Cooler: Noctua NH-D15 RAM: Adata XPG 2x8GB DDR3 (XMP: 2133MHz 10-11-11-30 CR2, custom: 2203MHz 10-11-10-26 CR1 tRFC:230 tREFI:14000) GPU: Asus GTX 1070 Dual (Super Jetstream vbios, +70(2025-2088MHz)/+400(8.8Gbps)) SSD: Samsung 840 Pro 256GB (main boot drive), Transcend SSD370 128GB PSU: Seasonic X-660 80+ Gold Case: Antec P110 Silent, 5 intakes 1 exhaust Monitor: AOC G2460PF 1080p 144Hz (150Hz max w/ DP, 121Hz max w/ HDMI) TN panel Keyboard: Logitech G610 Orion (Cherry MX Blue) with SteelSeries Apex M260 keycaps Mouse: BenQ Zowie FK1

 

Model: HP Omen 17 17-an110ca CPU: i7-8750H (0.125V core & cache, 50mV SA undervolt) GPU: GTX 1060 6GB Mobile (+80/+450, 1650MHz~1750MHz 0.78V~0.85V) RAM: 8+8GB DDR4-2400 18-17-17-39 2T Storage: HP EX920 1TB PCIe x4 M.2 SSD + Crucial MX500 1TB 2.5" SATA SSD, 128GB Toshiba PCIe x2 M.2 SSD (KBG30ZMV128G) gone cooking externally, 1TB Seagate 7200RPM 2.5" HDD (ST1000LM049-2GH172) left outside Monitor: 1080p 126Hz IPS G-sync

 

Desktop benching:

Cinebench R15 Single thread:168 Multi-thread: 833 

SuperPi (v1.5 from Techpowerup, PI value output) 16K: 0.100s 1M: 8.255s 32M: 7m 45.93s

Link to comment
Share on other sites

Link to post
Share on other sites

2 minutes ago, Jurrunio said:

It's the FIVR approach like on Skylake-X. Whatever you do on the outside will change things you want to keep unchanged.

Well judging by the name i guess the cpu controls the voltage, and judging by the "change things you want to keep unchanged" means that if you f around with voltage reference for FIVR then all the other voltages go up also demolishing stability

 

Though could throttlestop raise sa volt?

Link to comment
Share on other sites

Link to post
Share on other sites

2 minutes ago, Somerandomtechyboi said:

Though could throttlestop raise sa volt?

I doubt it can. The only exception to "the BIOS can't but TS can" is when the BIOS itself is too locked down to offer such option, say my laptop on Vcore. When it's a hardware or at best Intel ME lock, TS couldn't do anything about it.

CPU: i7-2600K 4751MHz 1.44V (software) --> 1.47V at the back of the socket Motherboard: Asrock Z77 Extreme4 (BCLK: 103.3MHz) CPU Cooler: Noctua NH-D15 RAM: Adata XPG 2x8GB DDR3 (XMP: 2133MHz 10-11-11-30 CR2, custom: 2203MHz 10-11-10-26 CR1 tRFC:230 tREFI:14000) GPU: Asus GTX 1070 Dual (Super Jetstream vbios, +70(2025-2088MHz)/+400(8.8Gbps)) SSD: Samsung 840 Pro 256GB (main boot drive), Transcend SSD370 128GB PSU: Seasonic X-660 80+ Gold Case: Antec P110 Silent, 5 intakes 1 exhaust Monitor: AOC G2460PF 1080p 144Hz (150Hz max w/ DP, 121Hz max w/ HDMI) TN panel Keyboard: Logitech G610 Orion (Cherry MX Blue) with SteelSeries Apex M260 keycaps Mouse: BenQ Zowie FK1

 

Model: HP Omen 17 17-an110ca CPU: i7-8750H (0.125V core & cache, 50mV SA undervolt) GPU: GTX 1060 6GB Mobile (+80/+450, 1650MHz~1750MHz 0.78V~0.85V) RAM: 8+8GB DDR4-2400 18-17-17-39 2T Storage: HP EX920 1TB PCIe x4 M.2 SSD + Crucial MX500 1TB 2.5" SATA SSD, 128GB Toshiba PCIe x2 M.2 SSD (KBG30ZMV128G) gone cooking externally, 1TB Seagate 7200RPM 2.5" HDD (ST1000LM049-2GH172) left outside Monitor: 1080p 126Hz IPS G-sync

 

Desktop benching:

Cinebench R15 Single thread:168 Multi-thread: 833 

SuperPi (v1.5 from Techpowerup, PI value output) 16K: 0.100s 1M: 8.255s 32M: 7m 45.93s

Link to comment
Share on other sites

Link to post
Share on other sites

9 minutes ago, Jurrunio said:

I doubt it can. The only exception to "the BIOS can't but TS can" is when the BIOS itself is too locked down to offer such option, say my laptop on Vcore. When it's a hardware or at best Intel ME lock, TS couldn't do anything about it.

Guess the only way to do this kind of voltmodding bullcrap is to start screwing around in the microcode if its just an arbitrary lock and not a hardware lock

Link to comment
Share on other sites

Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now

×