Took a peek at my PSU....and yes, I do have a 6+2 pin free, and easily accessible too. Looks like this is a no-brainer.. ;)
Question is which one to choose, whether to go for highest possible boost clock (something like the Asus ROG Strix) or a more moderately priced card.
Cheapest would be a EVGA 1070 SC Gaming ACX 3.0 (1594-1784MHz) at ~$550 over here.
The Asus ROG Strix Gaming (1657-1860MHz) would set me back ~$686
Must dig out some reviews.
Could wait and see if prices fall....that game is hard to win at.

. . My experience is that the higher end versions are more for gaming performance and can probably make a difference if you are a gamer. But for SETI the basic versions perform just about as well as their more expensive stablemates and give more bang for your buck! With prices the way they are that can be a good saving.

Stephen

:)

I agree with Stephen's assessment of the value of gaming oriented cards with higher clocks than the standard cards. The performance of SETI applications is more dependent on the number of shaders or compute units or CUDA cores, or whatever you want to call them than on clock frequencies. So buy the card that you can afford with the highest count of compute units even if the clocks are default standard design. GPU Boost 3.0 will boost the standard clocks much higher anyway on the standard designs just as well as on the gaming cards. This all applies if the card is just for compute. If the card is going to do double duty as a gaming card, then just change the evaluation to suit.

---

Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)

Loosing Server status page

---

This is odd. The SSP locks up often enough, but it doesn't usually just take a powder.
Meow?

---

"Freedom is just Chaos, with better lighting." Alan Dean Foster

Yup, the SSP has gone for a proper hike this time - just headers & footers and no content. Nothing to be done until the tyre kicker in chief gets into the lab in a few hours time.

---

Bob Smith
Member of Seti PIPPS (Pluto is a Planet Protest Society)
Somewhere in the (un)known Universe?

Well, so far it doesn't seem to have affected work flow here.
(Knocks on wood.)

---

"Freedom is just Chaos, with better lighting." Alan Dean Foster

SSP is back, but backup server is about 7 hours behind main....

---

Bob Smith
Member of Seti PIPPS (Pluto is a Planet Protest Society)
Somewhere in the (un)known Universe?

SSP is back, but backup server is about 7 hours behind main....

And 'ready to send' is about 200K down - so it probably wasn't keeping up while invisible.

SSP is back, but backup server is about 7 hours behind main....

And 'ready to send' is about 200K down - so it probably wasn't keeping up while invisible.

The sah assimilator processes are all down so something is wrong.

I figure if I post this info that they will immediately come back up so people can post "you're wrong" :-) which is usually what happens when I notice anything weird in the server status page

It almost always works

---

The server, or posting about it?

The server, or posting about it?

posting about it fixes the problem, thereby making whoever posted about it appear to be "wrong" later in the day.

---

Linux laptop:
record uptime: 1511d 20h 19m (ended due to the power brick giving-up)

SSP is back, but backup server is about 7 hours behind main....

And 'ready to send' is about 200K down - so it probably wasn't keeping up while invisible.

The sah assimilator processes are all down so something is wrong.

I figure if I post this info that they will immediately come back up so people can post "you're wrong" :-) which is usually what happens when I notice anything weird in the server status page

. . That is the supernatural nature of these threads, as soon as you write about something it changes ...

. . I cannot get any work ... oh wait, I just got 54 new tasks ...

Stephen

:)

Just reported the last of my stash of APs on the main cruncher (orion). I've got Ryzen+ coming on Saturday.

It's been a good run on this Bulldozer.. but the party isn't over with it yet. I'll be moving the board+cpu+ram over to altair, making that the.... fourth hardware migration for that machine since I built it in 2007.

---

Linux laptop:
record uptime: 1511d 20h 19m (ended due to the power brick giving-up)

Just reported the last of my stash of APs on the main cruncher (orion). I've got Ryzen+ coming on Saturday.

It's been a good run on this Bulldozer.. but the party isn't over with it yet. I'll be moving the board+cpu+ram over to altair, making that the.... fourth hardware migration for that machine since I built it in 2007.

I just got my old Bulldozer upgraded to Ryzen+. Did my standalone Win7 box first as guinea pig for transition from Window to Ubuntu 18.04 LTS. Hardware installation was a piece of cake. Fought the software for a few days, or at least the correct BOINC gpu configuration.

Still have to upgrade this daily driver to Ubuntu and another Ryzen+. Both this one and the standalone cruncher were identical twins mostly. Going to twin them again. Makes it easier for diagnosing and rebuilding if necessary.

---

Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)

A lot of Server Status data still MIA on the Haveland graphs, and one that is there has flatlined.

---

Grant
Darwin NT

Well so far, so good.


Was literally a point-and-click install process, once I made the updated ISO using the Gigabyte utility. Had usb3.0 drivers, nvme drivers, and those two KB updates needed for nvme. Install was a breeze, and all the driver installs are done, too. Only one "unknown device" in Device Manager--the onboard wifi, which is 10-only, but I use gigabit with a cord, so... don't care about that.

Other notes: I didn't fully pay attention when a friend of mine asked with a newegg link "is this the RAM you're thinking about getting?" and totally didn't see that it was CL14. CL16 3200 was in the QVL list, but not 14. Didn't realize that until after the order was placed. Did some quick research and saw that people were able to get the CL14 ram to work on x370 boards as far back as August last year, so I figured it should work.

Got everything hooked up and pressed the power button and it took about 30 seconds for the board to initialize and figure out what the defaults should be, and then booted fine. Went into BIOS and updated the BIOS from a file on a flash drive, restarted, and then went and changed XMP from "disabled" to "enabled" and restarted again.

3200 at 14-14-14-34.




And I've heard there are clearance issues on these boards these days.. it was close, but this RAM works with the stock cooler.

---

Linux laptop:
record uptime: 1511d 20h 19m (ended due to the power brick giving-up)

Absolutely, the preferred DDR4 memory is Samsung B-die, single ranked, two stick, CL14 3200 Mhz kits as the most compatible memory with Ryzen. Next would be 4 stick kits and then any dual ranked stick kits. The new X470 motherboards and the new Ryzen+ processors are easy to get running at 3200 Mhz. Literally drop in and play. With a little work, the higher clocks up to around 3533 are doable also. 3600 Mhz is a challenge and only for light workloads. I haven't seen any success stories and 3600 Mhz memory with fully loaded computers like with what we do with distributed computing. I have seen performance actually go down with a heavily loaded computer and clocks higher than 3200 Mhz. Seems the inter-CCX transfer rates and timings impact the higher clocks. The sweet spot seems to be 3200 and tight timings. You should be able to tighten up your 3200 memory with the Rzyen DDR4 Calculator available at OCN.net forum. The link to the latest calculator is here

Either the G.Skill TridentZ or FlareX 3200 CL14 memory is the best with Ryzen.

---

Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)

Yeah, one of the things I found in the research is that ALL g.skill CL14 memory is exclusively Samsung B-Die. 16's are mostly Hynix.

Unknown if that little factoid exists for the other brands.

But yeah, it was pretty painless. That utility on the Gigabyte website for the board to slipstream the ISO was a breeze. Made this installation point-and-click with absolutely no fuss.

....well, there was some fuss, but that's because I was trying to be very specific in the configuration. I got a 960 Evo 500gb and the intent was for it to have OS and all of my games on it (which is why I went for 500 instead of 250). At first, I didn't pay any attention to it, but I realized that the SSD was done with GPT--which I cannot do because a few of the games are old and they refuse to run from a "disk" that is GPT'd. I figured that one out last year when I upgraded my array to a 4x4tb raid5. I had a few of those games on a 500gb partition, and they refused to work. Moved them to the 128gb MBR'd SSD and they are fine.

So I started the installation over again, did shift+F10 to pull up the console, pulled up diskpart, nuked the SSD and made it MBR, made a partition (and formatted it, for good measure). Went to the install screen.. it refused to because "EFI systems cannot be installed to MBR devices."

Did research, tried about 5 different things, couldn't get it to work. When you mash F11 or F12 for a boot menu, you have to select the version of the flash drive that does not start with "UEFI:[device name]". Followed MS's instructions for making a flash drive bootable (diskpart, make a partition on it, format NTFS, set it as "active"). Copied the contents of the bootable installer over to it.. "bootmgr not found." Put that flash drive in the laptop and took the HDD out, put in a DVD version of the installer, shift+f10 and did /fixmbr /fixboot ... didn't fix it.

Was about to give up, and I noticed that if I open the .iso file in 7zip, there is a 375kb file called 'bootmgr' right there in the root. Copied that to the flash drive, tried again... "windows is loading setup files..."

and that's how I got 7 to install from a flash drive to a MBR'd SSD.

---

Linux laptop:
record uptime: 1511d 20h 19m (ended due to the power brick giving-up)

An easier way to get around the UEFI deal it to go to Boot options in the BIOS and select CSM Windows compatibility mode. Each motherboard BIOS is different so not sure if the Gigabyte board has the equivalent as my ASUS BIOS'. I expect so since that feature is actually from the AGESA SMU firmware code in the BIOS.

---

Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)

yeah, i had it in the mode for mixed/legacy. Like I said, it was just really convoluted to avoid EFI boot while also keeping the SSD in MBR mode.

And then I almost lost it all a few hours later once I stripped the rig down of the old Bulldozer to put this board in. I hooked everything up, including the old SSD, and mashed down on F12 (boot device menu) to select the 960 Evo. It booted up just fine. Installed the GPU drivers, restarted.. and then there was a Boot Manager asking which Windows 7 do I want to boot.. all one choice in the list, and keyboard didn't like that screen, even on a usb2.0 port.

An hour of trying all kinds of tricks and techniques and I just kept making the situation worse.. I was juuuuuuuuuuust about to grumble, gripe and groan and grab the win10 ISO, and got one last idea.

"Other PCI Device ROM Priority": UEFI Only (default). Changed it to "Legacy Only", save & exit, board restarted... Windows began loading just fine, without a boot menu.

I haven't plugged the old SSD back in yet.. I made a Macrium Reflect image of it before shutting the Bulldozer rig down, so I need to just pull that SSD and wipe it in another machine so that Windows doesn't detect TWO drives with Windows installations on them and give me that boot menu crap again. Urg.



But it's all good now. I'm up to about 90% back to where I left off on the previous setup. Just 5 or so more things to install or tweak/adjust and I'm 100% back on smooth sailing.

---

Linux laptop:
record uptime: 1511d 20h 19m (ended due to the power brick giving-up)