the multiplier is being limited at x17
the multiplier is being limited at x17
I'm adjusting the settings on my i7-920 to 166MHz. The motherboard model is SERVER_MB ASUS|P6T WS PRO X58 R. I've configured all the required BIOS settings, but CPU-Z keeps showing the multiplier capped at 17x (2819MHz). When "CPU Ratio Setting" is changed to 21x in BIOS, it still stays at 17x during stress tests. The bus speed appears correct at 166MHz, and Windows 10 power settings are set to high performance. I've also tried disabling Turbo, but it doesn't make a difference. What else might I be missing? I also wonder why CPU-Z displays a range of 12-22 in the parentheses even though I've specified a value.
Well, only the 1066Mhz setting seems manageable, so that should be okay to set up the CPU with. Just keep that in mind. I don't think it would push the CPU multiplier either.
For a 191Mhz BCLK, there are some unusual stable points to discover. Back then I remember some folks running 220Mhz and achieving around 4.4Ghz, I'm sure they had to test several CPUs before finding one that could handle it.
Running my memory at 1.65 volts was the final blow to my system. It was a gentle failure—triple channel just won't work—and the whole setup would only hold up to 8GB at once, though all slots still functioned properly.
It seems to be the standard range for the CPU only. The initial part should show the actual current speed. Have you checked if testing with BIOS settings of 17 and 21 gives a different result? If performance varies, CPU-Z might be overestimating. Does Windows match the 2.82Ghz claim? Are there other programs that also display the same speed?
I'm not entirely confident. My X58 board didn't have any issues, but it wasn't a 'professional' model. Since it's still aftermarket, I don't think there would be any major limitations.
I previously ran 161x23 with an i7-950 at 3.7 Ghz and undervolted it. My main issue was always making the memory run faster than 1600Mhz (that was quite fast back then).
I still think it might be a software problem, but I can't figure out what. Perhaps try a BIOS reset and only adjust the BCLK and memory speeds to see if that helps.
Certainly. Here is the rewritten version:
The BCLK on X58 is directly linked to memory speed through a multiplier configuration. For example, 133Mhz multiplied by 6 equals DDR3 800Mhz; 133Mhz times 8 equals DDR3 1066Mhz; and so on. I previously used 133Mhz with a factor of 12 to reach DDR3 1600Mhz, which was the maximum achievable at that time (around $300 for memory). If you were using 1333Mhz at 166Mhz, it would likely still function, depending on the RAM type. DDR3-2400, 2666Mhz and beyond were available later. This setup represented the initial DDR3 platform.
I didn't think about the RAM, could that be the issue? This is what I have:
https://imgur.com/G4d7dUU
View: https://i.imgur.com/G4d7dUU.png
These are my configuration details if you wish to experiment. 920 DO stepping, motherboard is a P6TDelux although settings might be comparable. RAM consists of three 4GB 1600 units; I had to reduce the frequency for stability at 1.5v. Likely could increase speed by raising voltages up to 1.65v, but it has worked well for many years. I'm using a Noctua D12 cooler inside an Antec 302 case. Apparently Bloomfield prefers odd multipliers, so I applied 21*191 to achieve the desired value.
Speedstep is disabled.
CPU Ratio: 21
BCLK Frequency: 191
PCIe Frequency: 100
DRAM Frequency: 1149
UCLK Frequency: 2298
QPI Link: AUTO
__
CPU Voltage: 1.27
CPU PLL: 1.88
QPI/DRAM: 1.27
DRAM Voltage: 1.5v
C1E Support: ENABLED
HT: ENABLED
A20M: DISABLED
Intel Cstate: DISABLED
Well, only the 1066Mhz setting seems manageable, so that should be acceptable when configuring the CPU. Just keep that in mind. I don't think it would push the CPU multiplier to any extreme.
For a 191Mhz BCLK, there are some unusual stable points to discover. Back then, I remember some users ran at 220Mhz and achieved around 4.4Ghz, which suggests they had to test several CPUs before finding one that could handle it.
Running my memory at 1.65 volts was the final factor that ended up damaging my system. It was a gradual failure—triple channel just wouldn't cooperate, and the whole setup would only support up to 8GB at once (all slots were functional too).
I've seen that running 1.65 works, yet I'm aware of Intel's specifications where 1.5 was the standard. Your experience makes sense to me. When this platform launched, DDR3 1.65 was the only option for a while. Those were unusual times.