Originally Posted by Sin0822
Hey so I actually did take this up the chain, things just take time, there is a new BIOS F4R which fixes the throttling issue for some CPUs but not all, that BIOS will come later. They said as they test out fully with other CPUs they will be added. So it might be a stepping thing as to which are solved and which are not, try it see if it is solved. I don't have a retail CPU, just the ES C1 stepping and I just tested it and there is no throttling i can make you a video if it inspires confidence if it doesn't work for you.
Now to answer some of the other complaints:
1. On X79 and Z77 and Z68 and P67, tRC is calculated by the chipset and cannot be higher than tRAS + tRP, that is why they don't let you take it higher, and if you do set it higher then the chipset will just default to tRAS+tRP just like if you take a setting too low. Intel does these things to simplify the BIOS, I asked the engineer, he said leave it on auto and it will goto the max it can.
2. Now about the RST rom, they said Intel wanted it this way as of recently, because they wanted to add smart response to X79 and thus a new ROM. However the engineer said they are working on adding both ROMs to the UEFI, this seems like a bigger issue for them than the throttling from the way things sounded, they said a new BIOS will be out soon.
First off, thanks for pushing this matter. Unfortunately, I just tested the F4r UEFI BIOS with a brand new fresh install of Windows 8 x64 and the throttling is still there irrespective of any settings changes in the UEFI BIOS
. I have a retail C2 chip (two of them in fact and have tested with two retail X79S-UP5 boards). If it will expedite resolving this matter, maybe I can provide one of the CPUs to Gigabyte on a loan? I am not sure how that could be arranged but I would consider the option as long as they turn around reasonably quickly. What type of LinX numbers do you get at a 4.4GHz overclock and at 4.6GHz overclock with your ES C1 chip?
As for the other two issues you addressed...
1. I'm still not convinced that this is the case. What is the justification for allowing manual control up to 32 and does that imply that the settings currently available have no effect what so ever? Why not simply allow us to set the value to whatever we want and apply that? It seems counter intuitive that this setting even exists in the UEFI BIOS if what you are saying is true (I have my serious doubts as to the accuracy of the information).
2. The F4r BIOS you posted added the ability to go back and forth between RSTe and IRST. This is a good step forward as I can at least now decide whether I want to run mainstream IRST or RSTe. The setting seems to work correctly. As for running both RSTe and SCU option ROMs at the same time, I think that is a separate problem from the one where I and a few others complained about the elimination of the RSTe option ROM (which as I said is back in the UEFI BIOS you posted). I am concerned that the message conveyed to me by support about there not being enough shadow memory to run two ROMs at the same time is actually correct, in which case my point about this being an oversight stands.
So, in all, F4r adds the ability to go back and forth between IRST and RSTe option ROMs, but it does not fix throttling or any other issues.
Originally Posted by Blameless
That's for the updates dejanh and Sin0822. I'll check out the new BIOS when I can get a chance.
To be honest, I'm not terribly concerned about the RAID ROM issue; I generally only have one or two boot volumes, and I do not need to put them on the SAS ports, so I'm ok with configuring those in the OS. Still, if there is a possible work around it would be nice.
Regarding tRC, I assume you mean IMC and not chipset, as the chipset shouldn't touch the memory. I have to admit, I'm still a little skeptical about what this engineer has stated. There are some X79 board that seem to allow tweaking increase of this timing past tRP + tRAS (evga boards, for example), and many XMP profiles attempt to set radically higher tRCs (not that XMP values are always optimal, but not off of them are worthless). Still, there are boards that seem to OC even large amounts of memory without issue, that do not have manual tRC control.
So, your engineer may well be right.
I think I'll see if I can under clock my memory and run very tight timings, and see if the max manual value (32) is recognized or if it makes any difference.
Those boards (like older and current Rampage Extreme boards) typically automatically set an unusually high value for tRC and are therefore capable of pushing some serious memory clocks with large number of DIMMs populated. As I said above, if what the FW engineer has said about tRC is actually true, then what is the point of this setting at all and why are we allowed to set it up to 32, and does that mean that this is yet another feature which exists in the UI and does nothing on the back end?