171.67.108.11 & 171.67.108.21 (old G8x/G9x GPUs are down)

[VijayPande]

We hoped this server would be back up by now, but there appears to be some hardware issue. The sysadmins are looking into it.

[VijayPande]

UPDATE: The machine is going through a long fsck, which takes a while since they have such large RAID's.

[toTOW]

Still both down on the server status page

[VijayPande]

fsck can take a very long time with huge RAID arrays.

[bruce]

"a very long time" is measured in days, not hours.

[Napoleon]

Though I agree with the rest of you that core 11 is likely on the way out, I would hope that whenever that does happen it would be accompanied by an announcement of some kind.

As the blog post I linked to in my earlier post suggests, PG already attempted to shut down core_11 work back in December and introduce core_15 work to pre-Fermi GPUs. Old G8x/G9x GPUs turned out to be problematic, some were OK with the change, others were not. I had the "privilege" to compare P762x WUs (all with core_15 v2.25) side by side on three GPUs (GPU_MEMTEST_ERROR topic):

1. GT430 card
2. 9400GT card
3. 1st generation ION: essentially an integrated 9400GT, using a slice of the motherboard memory

As you might guess, ION failed right away, whereas the "real" 9400GT (apparently) returned valid results. Might give you an idea how flaky the G8x/G9x thingies are when you try to introduce something newer than core_11 to them. If I had to guess, I'd say ????GT(X) cards might be on their way to blacklist when core_11 finally is done for, yet I suspect GT(X) 2?? cards are a different matter.

But I tend to speculate too much. The fact at the moment seems to be that we'll be seeing some more core_11 work once the old servers have gotten their share of Tender Loving Care.

Enjoy the weekend, folks.

[KroontjesPen]

Hopefully this fsck works and the servers coming up and running again.
Maybe the collection of done WU's can picket-up by other servers.

The pc where the GeForce 9600GT is in, works mainly for F@H.
If the support for the 9600GT is stopping I shall have to shutdown that pc and a SMP:4 client is coming down with that.
Because of the 3e rescission here I can't effort an upgrade to that system now.

Is the drop in completed WU's by this event notable?

[Napoleon]

Back in December...

I'll also look to see if we can get some of the old projects that use core_11 back up, but those were pretty much finished, although a bit more statistics is never a bad thing.

[farmpuma]

Good to know it's a server issue and not the core_11 EoL. I thought it was probably the server when my G92 wasn't even able to get a core_15 work unit. Is anyone getting core_15 WUs?

[P5-133XL]

Yes, core_15's are being given out. I have several GTX 460's that are getting p807x that use core_15

[toTOW]

It's been a long time since I saw the last core_15 WU on G8x/G9x cards ...

And we already know that OpenCL cores won't work on these old chips which don't support more than OpenCL 1.0

[djay86]

Is there a way to receive core_15 work assignment on GT200 cards anyway?

[Sliced]

Ah, I thought it was only me.
My 9800GTX+ has stopped folding as there is no assignment server.
I really hope they get this fixed and don't stop the older cards from folding as it's my main folding resource.
If not then I will just have to take it out and put it back in a box until I find another use for it.

[Joe_H]

As of about 8 AM PST, one of the VSP07 servers is back online, 171.67.108.21, and is accepting returns and giving out new assignments. 171.67.108.11 is no longer down, but is not yet accepting connections. I assume that means it needs more work to get it back online, but don't have any idea how long that will take.

[djay86]

Nice to see my GPU back to folding, it wasn't that long to repair this server.