Bug in B85M-ITX BIOS

[srr2]

I've just finished building a number of these and H87M Pro4 boards. All of them have had i5-4570S processors. I'm pretty sure that there is a bug in the BIOS (1.4) of the B85M-ITX involving setting the full processor multiplier. Comparing them side-by-side, the H87M, under full load, runs the processor at 3.6GHz on all cores. That's as it should be. (Naturally, that's how the BIOS is set). The B85M-ITX WITH THE SAME BIOS SETTINGS will not run all four cores faster than 3.2GHz simultaneously. There's no reason this should differ between boards. Needless to say, I've tried all possible settings to get this to work right, including setting manual ratios of 36-36-36-36. It will NOT achieve this on the processor. This is confirmed by both CPUID and actual benchmarks. CPUBenchmark of the H87M comes in at an astonishing 7846, while the same processor on the B85M is stuck at around 7100. This is consistent with the 3.2GHz limit. What actually happens despite the settings is that the processor behaves as if the cores are set for 36-35-34-32. I've simulated this loading with Prime95 and that's the best you can get out of it.

Could someone from ASRock please look into this and get the BIOS fixed?

[parsec]

Re: Bug in B85M-ITX BIOS

I've just finished building a number of these and H87M Pro4 boards. All of them have had i5-4750S processors. I'm pretty sure that there is a bug in the BIOS (1.4) of the B85M-ITX involving setting the full processor multiplier. Comparing them side-by-side, the H87M, under full load, runs the processor at 3.6GHz on all cores. That's as it should be. (Naturally, that's how the BIOS is set). The B85M-ITX WITH THE SAME BIOS SETTINGS will not run all four cores faster than 3.2GHz simultaneously. There's no reason this should differ between boards. Needless to say, I've tried all possible settings to get this to work right, including setting manual ratios of 36-36-36-36. It will NOT achieve this on the processor. This is confirmed by both CPUID and actual benchmarks. CPUBenchmark of the H87M comes in at an astonishing 7846, while the same processor on the B85M is stuck at around 7100. This is consistent with the 3.2GHz limit. What actually happens despite the settings is that the processor behaves as if the cores are set for 36-35-34-32. I've simulated this loading with Prime95 and that's the best you can get out of it.

Could someone from ASRock please look into this and get the BIOS fixed?

No bug in the B85M-ITX board's BIOS, and you've actually partially answered your own question. The main reason is the B85 chipset does not support over clocking. BTW, an i5-4750S processor does not exist, it is the i5-4570S, or i5-4670S.

As you said, the i5-4570S not only behaves as if it operates in Turbo mode at 36-35-34-32, it does just that. Since the B85 cannot OC, and therefore does not have an unlimited Turbo time limit, it will not stay at those core speeds, and then change to 32 on all cores. The "rules" of Turbo only allows the Turbo speeds to be sustained for several seconds, using the standard Turbo time limit value.

But what about the "non-Z OC" feature? That only works with 'K' processors.

Frankly, you're lucky the i5-4570S will operate at 3.2GHz. If the BIOS obeyed all the Turbo and OC rules, it would clock down to 2.9GHz, its base operating frequency. Something in the 1.40 BIOS is giving you the extra MHz, or the extra was available in previous versions too, or the few extra "bins" some of the non-K processors are supposed to have is what you are getting.

You're lucky that the H87 board allows the CPU to operate at 3.6GHz on all cores, since that chipset also does not support over clocking. Only the Z87 chipset supports over clocking, they don't call the feature "non-Z OC" for no reason.

Also, Intel has started to crack down on the mobo manufactures that are offering the "non-Z OC" feature, since they do not sanction that feature. I'd keep a copy of the 1.40 BIOS, since it will no longer be available after it is removed, which apparently will happen.

[srr2]

Re: Bug in B85M-ITX BIOS

Okay, while I appreciate the time you took to reply to this, after spending over an hour poring over Intel's own documentation on this subject I have respectively to disagree. There's nothing off-spec about operating all four cores at 3.6GHz for this processor, likewise, I can find nothing that confirms any difference between B85 and H87 chipsets in this respect. If there is such documentation, I failed to find it and would appreciate a specific reference to it if you have one.

So for the moment, I'll stand by my original post.

[parsec]

Re: Bug in B85M-ITX BIOS

I'm not concerned with right and wrong here, I'm just trying to describe why the limits you are seeing are there.

While Intel supposedly supports over clocking on some processors and chipsets, they don't document it at all, as you know. Which is why I say supposedly.

It could be said that there is nothing on-spec about operating all four cores at 3.6GHz for this processor.

My true point is this: The Haswell platform officially supports over clocking only on 'K' processors and the Z87 chipset. Any over clocking capability available with other Haswell processors and chipsets is provided by mother board manufactures by bending or breaking the rules as defined by Intel.

Since the OC capability you get on the H87 board with the i5-4570S is beyond the specifications for the chipset, how can we say that another chipset or its BIOS has a bug if it also does not operate beyond its specifications? The B85 chipset may have a limitation the H87 does not regarding the potential to OC processors, but neither is required by their specs to operate as you described.

[srr2]

Re: Bug in B85M-ITX BIOS

Let's get something straight here. THIS IS NOT OVERCLOCKING. The 3.6GHz is straight off the Intel specification for the chip. All four cores CAN run at 3.6GHz, and there's nothing in any Intel specification for chip or chipset that prevents/discourages/inhibits this IN-SPEC operation of the chip. The only limitations in running all cores at 3.6GHz are essentially thermal and ad-hoc. If you keep the chip cool enough, 4 x 3.6GHz is fully supported. Period.

You are confusing this with the "Non-Z OC" that at least ASRock, Asus, MSI, and Gigabyte (and probably others) have been featuring lately. That applies ONLY TO xxxxK series processors. Those are the only ones that are going to get the microcode update. The non-K processors will remain unmolested, since they don't have unlocked multipliers in the first place.

Therefore I stand by my original post: THERE IS A BUG IN THE BIOS FOR THE B85M-ITX BOARD.

[imo76]

Re: Bug in B85M-ITX BIOS

Let's get something straight here. THIS IS NOT OVERCLOCKING. The 3.6GHz is straight off the Intel specification for the chip. All four cores CAN run at 3.6GHz, and there's nothing in any Intel specification for chip or chipset that prevents/discourages/inhibits this IN-SPEC operation of the chip. The only limitations in running all cores at 3.6GHz are essentially thermal and ad-hoc. If you keep the chip cool enough, 4 x 3.6GHz is fully supported. Period.

You are confusing this with the "Non-Z OC" that at least ASRock, Asus, MSI, and Gigabyte (and probably others) have been featuring lately. That applies ONLY TO xxxxK series processors. Those are the only ones that are going to get the microcode update. The non-K processors will remain unmolested, since they don't have unlocked multipliers in the first place.

Therefore I stand by my original post: THERE IS A BUG IN THE BIOS FOR THE B85M-ITX BOARD.

From intel spec, 4570S has the following turbo boost frequencies: (1C/2C/3C/4C) 3.6-3.5-3.3-3.2

you can check by youself:



so what you are getting from your B85M is the intel stock frequencies. If anything abnormal, it is about your H87 motherboard used in the comparison making your processor work out of spec.

[parsec]

Re: Bug in B85M-ITX BIOS

From intel spec, 4570S has the following turbo boost frequencies: (1C/2C/3C/4C) 3.6-3.5-3.3-3.2

you can check by youself:



so what you are getting from your B85M is the intel stock frequencies. If anything abnormal, it is about your H87 motherboard used in the comparison making your processor work out of spec.

Thanks for posting the link to the Turbo boost table, but I doubt that will convince srr2.

I agree with you about the H87 board compared to the B85M board.