images.jpgI am not your standard reviewer and you will not find the stock Intel die images and graphs with the performance numbers. No need to tell you how wow the new heat sink is or go over already known TDP quotes. No need to bore yourself with six pages of rehashed info found on every other site. I have tested and compiled a lot of information to the everything leading up to B1 stepping in both Xeon and X980. I will not be showing any Xeon screen shots or performance numbers for embargo date.
Spare yourself the grief of falling victim to buying an A0 stepping. There are people are passing dead A0 chips around on the resale,. These chip will die when the wind changes, at stock speeds and in shipping... ES chips are not for sale parts, buying and selling of these parts are being cracked down on by Intel. Do not expect return of money or RMA warranty from any ES part.
With the dud B0 stepping right after a promising A0 stepping people are wondering what does the future hold. As I am about to show, there are 6Ghz B1 stepping parts out there. People are a bit scared after seeing B0 refinements took a hit on top clock speeds of those parts. No need to seek a B0 unless you get it for free.
http://valid.canardpc.com/show_oc.php?id=1069768
Granted it is not all cores but it is the part operating past 6Ghz on cascade and this is promising.
Q3QP Q3FE
A0 has no thermal protection and high leakage, this is a great combo for overclocking but they have a weaker memory controller and other early revision issues and lack of needed changes that went into B0. My A0 died the first day.
Q3VG
B0 is pretty good, not so great for overclocking past 6Ghz, more like 5.6Ghz is a good target. You will find memory clocking to be inferior to B1 stepping. I was not able to kill a B0 but I do think the chip is hurting.
Q4EG Q4HH - Current
B1 is pretty good overclocking and has noticeably stronger MC than A0 and B0. B1 is also very durable. Motherboard will die before you can kill it with vcore under controlled conditions beyond -100c.
I am missing a few QSPEC's but you can follow the pattern.
Some boards needed modification in order to post with Gulftown. All needed bios update. There is a difference with the Xeon variant and dual QPI working on existing boards. There was another microcode and firmware update between B0 and B1 to post on the same board. Here are some notes from a few of the top boards.
Gigabyte X58 UD7 needed hard mod, hicookie posted the info in the UD7 thread. Both X980 and Xeon now work great on this board with the exception to the OCP protection kicking in around 5.5Ghz with all cores loaded. Memory overclocking is very strong and performance was exceptional. Mod difficulty of 7, the SMT resistor value must be soldered onto. I did have a bios update brick my working first UD7 board. staying under the OCP the board have good stability in all benches. This board needs to be walked up in speed for best results with the OCP and an issue with too high of vcore on first post. Meaning if the system first up at 1.5vcore you can step to 1.7 and higher, make changes to number of cores, enable HT. Once you go past ~1.6v the board will ignore any changes to active cores and HT as well as anything that results in strap change will force you to start over from bios reset to make any changes at all. What you see is C1 stuck on the post LED, it will cycle C1 quickly on fail.
Once again, reset bios and load profile with all the changes needed to cores, HT, and straps. Once you go too far these cannot be changes once at top speed without resetting bios and starting over with first known good.
Asus Rampage II and III Extreme just needed a bios update for both X980 and Xeon versions to fire up, overclocking is straight forward and very strong. No OCP issues. This board also has strong memory clocking and performance was good overall. I will be streaming live the R3E submerged in a new custom case Mike and I built in the next few days. I had to build sinks for this board like a few other people who got the board early.
EVGA Classified needed a bios update for X980 to work, a hard mod is needed for Xeon to work, mod difficulty of 3. You need to 0k two resistors in order for Xeon to function.
Intel X58SO works with just a bios update, no issues with minimal overclocking function of all the boards listed. Don't underestimate this rock solid board as I reached 5.67Ghz pre launch of Bloomfield fairly easy.
MSI Eclipse needs a bios update. If this bios is not posted yet, hit me up if you need it.
After many hours of sub zero testing on the boards above I found the Asus R2E the best for bios and functionality so far and I plan to re run the R3E tonight live.
You will find the memory controller hit due to cache latency hit to be minimal as programs like favor larger cache will accelerate on this architecture.
Chances are if you use an XMP profile it will work on B1 stepping.
Overclocking yields higher clock speeds than 45nm counterparts but you are still dealing with six cores and twelve threads. Draw from wall was 630 watts @ 5.5Ghz when OCP kicked in every time, OCP on the UD7 is not an issue for anything less than a cascade.
I am sure most board issues that require mods will be fixed quickly now that launch day has passed and chips will be entering the market.
Onto the benches broken down to each board in a separate thread. Hopefully you found some of this information useful.
Reply With Quote
Almost all new high end MBs requite a hard mod to boot with 32nm Xeons?
Originally Posted by FUGGER
Bookmarks