22
Replies
OK, I should have said "It's build WEEK"
Here seems to be the relevant part of the manual. To me it seems pcie3_3 and pcie3_5 are the options. _3 shares bandwith with the onboard m.2 port which I'll be using and the u.2 ports which I wont be. pcie_5 shares with another u.2 port I won't be using. So, those two slots, is there any preference which to each one going where? I think maybe slot 5 for the hyper m.2 card, since that will be in use constantly, then the TB card in _3?
I am confused by this data from the instruction sheet. It shows that even with the M.2 enabled that you should be getting x16, x8, and x8 on the first three slots. That is more than the required x4 for the Thunderbolt and the Hyper M.2 card require so I do not see why you could not just use slots 1, 3, and 4. (Slot 2 is just an x1 slot that is generally hidden by the double width GPU) Maybe Ronin or someone else can straighten me out.RoninEdits
I finally found out why slot 3 is no good it is a PCIe 2.0 slot only, not suitable for Thunderbolt or Hyper M.2 caqrd Below they are showing Slots 1, 3, 4 I think? So you indeed would have to try
Actually only Slot1 and Slot 5 (PCIex16_4) are not shared by anything. I am sure glad I got the workstation x99 as it does not have all these if's, and's an but's.
*1:PCIe x16_3 shares bandwidth with M.2 and U.2_2. It runs at x16 mode by default. PCIe X16_5 shares bandwidth with U.2_1. It runs at x4 mode by default with U.2_1 enabled.
*3: PCIe x16_2 shares bandwidth with USB3_34 and USB 3.1_EA34. It runs at x2 mode by default.
Sorry but I give up!
Ronin
Thanks for the update. My workstation board I can have two x16 GPU's, the motherboard M.2 x4 plus two M.2 adapters installed and all running at rated speeds. And actually the x99E-WS can run 4 GPU's at x16 with the PLX chips
This my 960 Pro M.2 1 TB on the left. Next to it is the 950 Pro 512 GB boot drive and below it is my 960 Pro 512 GB M.2. Both GTX 1060's are overclocked and running at PCIe 3 x16
OK build month! 
So, the rig was mostly built and I started running some items on there.
I ran 3DMark on it, and mostly it ran well and reported it to be very good. Except it rebooted a few times.
So I got out some stress test software. I started to run Prime95 on it. The default test Blended caused an immediate BSOD. Tried again, same deal. Tried the Small FFT and that ran thru. Then Blended again which caused a reboot. So, I ran the Windows built in “Memory Diagnostic Tool” and after a short time it reported ‘hardware’ issues. So, I got out MemTest86+, and ran it overnight that too reported bad memory and 252 errors. So, yesterday I RMA'd the memory. G.Skill has some crazy times on their website which end up totaling 10 days, I managed to get hold of someone via chat and he said he would expedite. They told me on the phone today the memory shipped so I may get it tomorrow. That wouldn't be too painful.
All in all, looking good, and I'm happy with it. It's super quiet, that's for sure.
First thing though, I was happily watching Bill Gehrke explain that 'you should not have much more than 70 processes running' which I have on my current machine (Win 7), but Win 10 has like 150 running with the initial install!! Yikes! So, I'm going to need some optimization. If anyone can point me to a good itemized reference, that would be great.
Also, somehow I think my bios got set to XMP on, and thus some test software was saying I had an overclock going. Which I didn't do anything to setup. So, does having XMP on constitute an overclock? And does that default to on?
Here's an a couple shots of my rig:
I've been up and running of course, but still tweaking it here and there. Was able to run PPBM many times, I got good at it. Even had a shortcut to Project Settings/General. 
This was my best run. Left the system just a tick down on this at 44ghz on all cores at 1.225v, VERY QUIET! I hadn't realized before that the Windows Power settings manipulated the CPU clock speeds. I wanted to keep my processor and GPU clocking down at idle and managed that well. I got 18's on the Disk I/O several times otherwise it was 19. This was C: (960evo) OS/Programs/Default Cache and previews, F: Project, Files and Export (960pro). Only the h.264 times seemed a bit off. Tried running it in different ways, but kept getting around 60-63. BTW: Saw some drop off in speed having SnagIt open.
Thanks Bill, I really appreciate the feedback. I had the core clock at 44/45. Two of my cores 2 + 4 run about 3-5* hotter than the others with full load. So, I was trying some overclocking at the time to balance out the temps and thus ran the other cores faster. I was pulling in 301-305 on the disk i/o tests. I left it at 4.4ghz in the end so I could use the Asus recommended "sync all cores". On the 1070, I did have the memory clock raised just a bit. I did some overclocking on the clock speed and I was able to get +75->100 to run with the Heaven benchmark. In the end, the 1070 FTW2 is already clocked quickly and seems plenty fast, so I've left the 1070 at default settings for day to day work, but run XOC at startup to spin up the fans from their default 'off' (bummer you need to have the software running to speed up the fans. (unlike the computer BIOS (uefi)). Here's a screen shot of Crystal having run the two 960 (F: PRO, c: EVO) at the same time.
AdChoices