Prometheus/Andromeda, (Win 7/MAC, El Capitan 10.11.4), hybrid system has one GTX-750TI SC already installed. I just picked up a second. I'd like to have BOINC make use of both cards for the WOW Event.

What's the EASIEST way of doing this??? Do I add a line to the app_config.xml? Do I have to go to app_info.xml; which I'm uncomfortable working on/with... I'd rather do app_config.xml if possible.

I know that it has to do with "Use All GPUs" but cannot remember how to add the proper line...

Any and all help is greatly appreciated. :-) Thanks in advance. :-)

(I will primarily be running BOINC in MAC OS/on Andromeda... I'm using TBar's CUDA65 App.)


TL

---

TimeLord04
Have TARDIS, will travel...
Come along K-9!
Join Calm Chaos

If they are the same type of GPU then you should be able to just install it and it should run.

The only time you need to add a commandline is when you are using different generation GPUs.

If they are the same type of GPU then you should be able to just install it and it should run.

The only time you need to add a commandline is when you are using different generation GPUs.

Well that makes things easy... Thanks! :-)

They are identical models; both EVGA GTX-750TI SC, the original that is currently in place is the one I got from you - card works FANTASTIC btw. The second card is identical to the one I got from you; but, is just brand new. Do I need a Video Terminator installed at the DVI-I port on the second card, or will BOINC run without monitor, or terminator attached on the second card?

Thanks again. :-)


TL

---

TimeLord04
Have TARDIS, will travel...
Come along K-9!
Join Calm Chaos

Do I need a Video Terminator installed at the DVI-I port on the second card,

Nope.

---

Grant
Darwin NT

BAD NEWS!!!

The new card is physically different from the old card. The new card is missing physical cutouts in the circuit board and now won't fit the PCI-e socket without physical modification; which I have NO tools, nor talent to modify the card. The PCI-e circuit board on the new card is completely smooth to the touch on the edge; no "female" slots to accommodate the male projections of the PCI-e socket on the MOBO...

The old card VISIBLY has the two appropriate cutouts in the circuit board to match the PCI-e socket on the MOBO.

Anyone have a two year old EVGA GTX-750TI SC and want to swap for a new one???

I don't feel right returning a working card to Amazon just 'cause EVGA changed the physical design of the card...

PM me, please.

Thanks.


TL

---

TimeLord04
Have TARDIS, will travel...
Come along K-9!
Join Calm Chaos

The new card is physically different from the old card. The new card is missing physical cutouts in the circuit board and now won't fit the PCI-e socket without physical modification;

If it is a PCIE card, it will fit in a PCIe slot.
However there are *1, *4 *8 & *16 slots.
*16 are the ones used for video cards, and are the longest connectors.


Edit- looking at the image for your motherboard on Gigabytes web site, it shows 2 PCIe*16 connectors (one blue, one orange) and 2 PCI connectors. There are also 3 PCIe*1 one slots, and no the board won't plug in to those.

---

Grant
Darwin NT

You are not making any sense.

Your mobo has two full size PCI-e slots. First is blue, second is orange. One card goes into blue PCI-e slot, one card goes into orange PCI-e slot.

I have intel P45 chipset mobo with two full size pci-e slots and I had not trouble installing two GTX750Ti SC.

---

.....

---

BAD NEWS!!!

The new card is physically different from the old card. The new card is missing physical cutouts in the circuit board and now won't fit the PCI-e socket without physical modification; which I have NO tools, nor talent to modify the card. The PCI-e circuit board on the new card is completely smooth to the touch on the edge; no "female" slots to accommodate the male projections of the PCI-e socket on the MOBO...

The old card VISIBLY has the two appropriate cutouts in the circuit board to match the PCI-e socket on the MOBO.

Anyone have a two year old EVGA GTX-750TI SC and want to swap for a new one???

I don't feel right returning a working card to Amazon just 'cause EVGA changed the physical design of the card...

PM me, please.

Thanks.


TL

...I'm using TBar's CUDA65 App...

Hmmm, my recent EVGA cards have shipped with a rubber PCIe cover in place. You do need to remove that before installing the card. There are also numerous plastic stickies on the card that need to be removed. I have all my cards connected to the same monitor, the monitor has 3 connectors. Just make sure the monitor is set to the card you want to use as the main display.
There is a better App than the old cuda65 App over at C.A., you would probably be better off with the newer cuda75 App. Just make sure to go to System Preferences/CUDA and install the latest driver before using the new App.

OK - To be perfectly clear.

After discussing the matter with Zombu2, this new card I have is DEFECTIVE.

The card is missing the Dremel cutouts in the circuit board slot that inserts into the MOBO PCI-e 16 slot Number 2. The card I got from Zalster has the cutouts/female slots that match and meetup with the male protrusions of the MOBO PCI-e socket/slot.

There is a front cutout and a rear cutout of the circuit board on the card Zalster gave to me for $50 that the new card is completely missing. So, the new card WON'T seat into the MOBO slot.

I have just completed an RMA Ticket with Amazon. They are shipping me a replacement EVGA GTX-750TI SC card. I will be shipping the defective unit back tomorrow.

I wish I could upload pictures here; but, I'm not willing to move these from my iPhone to Photobucket... You have to understand, the circuit board that goes into the PCI-e socket of the MOBO is missing the necessary cutouts to seat the card.


TL

---

TimeLord04
Have TARDIS, will travel...
Come along K-9!
Join Calm Chaos

You have to understand, the circuit board that goes into the PCI-e socket of the MOBO is missing the necessary cutouts to seat the card.
TL

Ok. Got it.

Can I just say that it is very very very very weird.

---

Wow, I agree, this is a first for me as well. I totally and completely believe you, but as I've never seen what you are describing, would there be any chance that you could post a pic of the bad card, or even both of them just to show the difference? It would be neat to see, not that it is a neat situation for you of course... :-/

---

OK - Addendum... I'M A FRAKKUP!!!! I'M BLOODY STUPID, AND I FREELY ADMIT IT!!!!!

OK - Zombu2 pointed out that the damned card is shipped with a Rubber Protector over the circuit board! I didn't even pick up that THAT is what it was that was keeping the card from seating!!!

Rubber Protector NOW REMOVED!!! Card is seated, card is functioning. Currently crunching Einstein. BOINC now recognizes BOTH cards, and the one Zalster gave me is GPU Device 0, the new card is GPU Device 1.

So, I'm FINALLY crunching 4 Units at a time, 2 per card!!!

<<----- IDIOT!!!!! ;-)

OK - You can all laugh yourselves sick that I was THIS STUPID!!!!!

FRAKK ME!!!!! LOL


TL

---

TimeLord04
Have TARDIS, will travel...
Come along K-9!
Join Calm Chaos

LoL

---

I came down with a bad case of i don't give a crap

BAD NEWS!!!

The new card is physically different from the old card. The new card is missing physical cutouts in the circuit board and now won't fit the PCI-e socket without physical modification; which I have NO tools, nor talent to modify the card. The PCI-e circuit board on the new card is completely smooth to the touch on the edge; no "female" slots to accommodate the male projections of the PCI-e socket on the MOBO...

The old card VISIBLY has the two appropriate cutouts in the circuit board to match the PCI-e socket on the MOBO.

Anyone have a two year old EVGA GTX-750TI SC and want to swap for a new one???

I don't feel right returning a working card to Amazon just 'cause EVGA changed the physical design of the card...

PM me, please.


Thanks.


TL

...I'm using TBar's CUDA65 App...

Hmmm, my recent EVGA cards have shipped with a rubber PCIe cover in place. You do need to remove that before installing the card. There are also numerous plastic stickies on the card that need to be removed. I have all my cards connected to the same monitor, the monitor has 3 connectors. Just make sure the monitor is set to the card you want to use as the main display.
There is a better App than the old cuda65 App over at C.A., you would probably be better off with the newer cuda75 App. Just make sure to go to System Preferences/CUDA and install the latest driver before using the new App.

Missed this post; thank you for that info, I'll keep an eye out for this on FUTURE purchases as well...

<<--- AGAIN - FRAKKUP!!!!!

LOL

[EDIT:]

Card now crunching through Einstein cache. Once done, I will switch to SETI and see if both cards pickup there... They should, since they work on Einstein...

Then I'll register for the WOW Event.

Thanks again everyone.

OK - I think there's enough laughter now...

<<---- Still a FRAKKUP!!!!!

---

TimeLord04
Have TARDIS, will travel...
Come along K-9!
Join Calm Chaos

OK - I think there's enough laughter now...

It was a good-funny!
...and welcome to the 2x GTX 750 Ti club!

Are you using the NV_SoG app or Cuda50?

---

OK - I think there's enough laughter now...

It was a good-funny!
...and welcome to the 2x GTX 750 Ti club!

Are you using the NV_SoG app or Cuda50?

Currently, I'm using Cuda65 from TBar; but, he wants me to move up to his Cuda75 App... (MAC OS X, 10.11.4 on Andromeda. I'm on Alternate NVIDIA Driver 346.03.06f01, CUDA Driver 7.5.26)


TL

---

TimeLord04
Have TARDIS, will travel...
Come along K-9!
Join Calm Chaos

...and welcome to the 2x GTX 750 Ti club!
Are you using the NV_SoG app or Cuda50?

Currently, I'm using Cuda65 from TBar; but, he wants me to move up to his Cuda75 App... (MAC OS X, 10.11.4 on Andromeda. I'm on Alternate NVIDIA Driver 346.03.06f01, CUDA Driver 7.5.26)

Oops, I didn't see it was a Mac...but that explains a bit more the funny! ;-p

On CPU: are Guppis much faster than nonVLARs, and
on GPUs: are Guppis much slower than nonVLARs?
If so, I hope for you & other Macs in the Wow event that someone adapt's Mr. Kevvy's script to optimize the 2 queues.
Otherwise you'll be at a 5-10% disadvantage as compared to Windows or Linux with similar hardware. :-p

Cheers and best of luck to the Wow event 2016 Cancer team! [edit]intended as a playful banter[/e]
Rob ;-D

---

...and welcome to the 2x GTX 750 Ti club!
Are you using the NV_SoG app or Cuda50?

Currently, I'm using Cuda65 from TBar; but, he wants me to move up to his Cuda75 App... (MAC OS X, 10.11.4 on Andromeda. I'm on Alternate NVIDIA Driver 346.03.06f01, CUDA Driver 7.5.26)

Oops, I didn't see it was a Mac...but that explains a bit more the funny! ;-p

On CPU: are Guppis much faster than nonVLARs, and
on GPUs: are Guppis much slower than nonVLARs?
If so, I hope for you & other Macs in the Wow event that someone adapt's Mr. Kevvy's script to optimize the 2 queues.
Otherwise you'll be at a 5-10% disadvantage as compared to Windows or Linux with similar hardware. :-p

Cheers and best of luck to the Wow event 2016 Cancer team! [edit]intended as a playful banter[/e]
Rob ;-D

Rob,

Well, NOT ONLY is Prometheus/Andromeda a hybrid system; (Win 7 Pro x64 on one drive, and MAC OS X on another drive), but the Andromeda side is really a Hackintosh! :-O I've created kind of a "Frankenstein" out of old PC Hardware.

Prometheus is the Win 7 side of the system, and Andromeda is the MAC side... I'm holding tight to El Capitan 10.11.4 due to the fact that after Upgrading to 10.11.4 the Stock BOINC SETI OpenCL Apps were NOT functioning to true MAC Specifications, (while on 10.11.3 the OpenCL Apps were just fine...), so with MUCH help from TBar and others I Upgraded to TBar's CUDA65 App. Due to the fact that Andromeda is a Hackintosh, I'm hesitant to Upgrade to CUDA75; as originally TBar mentioned that Hackintoshes and true MAC Laptops still encounter issues with "Validation Inconclusives" and "Invalids" and even "Errors"... I'm afraid to Upgrade to CUDA75 as this might make things worse.

Other Apps, Games, and Programs on Andromeda have NO ISSUES; just BOINC SETI has issues. Einstein, (which ONLY sends CUDA55 work to Andromeda), works just fine. So, there's something not quite right with MAC OS X, the CUDA65/75 and OpenCL Apps, or both when running on Hack and MAC Laptops.

For my sanity, I'm sticking with what I have. It works as well as it can for me.


TL

---

TimeLord04
Have TARDIS, will travel...
Come along K-9!
Join Calm Chaos

Well, NOT ONLY is Prometheus/Andromeda a hybrid system; (Win 7 Pro x64 on one drive, and MAC OS X on another drive), but the Andromeda side is really a Hackintosh! :-O I've created kind of a "Frankenstein" out of old PC Hardware.

Prometheus is the Win 7 side of the system, and Andromeda is the MAC side... I'm holding tight to El Capitan 10.11.4 due to the fact that after Upgrading to 10.11.4 the Stock BOINC SETI OpenCL Apps were NOT functioning to true MAC Specifications, (while on 10.11.3 the OpenCL Apps were just fine...), so with MUCH help from TBar and others I Upgraded to TBar's CUDA65 App. Due to the fact that Andromeda is a Hackintosh, I'm hesitant to Upgrade to CUDA75; as originally TBar mentioned that Hackintoshes and true MAC Laptops still encounter issues with "Validation Inconclusives" and "Invalids" and even "Errors"... I'm afraid to Upgrade to CUDA75 as this might make things worse.

Other Apps, Games, and Programs on Andromeda have NO ISSUES; just BOINC SETI has issues. Einstein, (which ONLY sends CUDA55 work to Andromeda), works just fine. So, there's something not quite right with MAC OS X, the CUDA65/75 and OpenCL Apps, or both when running on Hack and MAC Laptops.

For my sanity, I'm sticking with what I have. It works as well as it can for me.

TL

???
Those "Validation Inconclusives" and "Invalids" and even "Errors" are when using the OpenCL App...Not the CUDA Baseline App. The Baseline App at C.A. works perfectly Fine as far as I know. You Do have to Install the Correct driver, which for Mavericks and above means the latest driver found in System Preferences/CUDA. The only problems I know of are caused by people updating the OS and Not updating to the driver that supports the New OS. There are a few people out there that have run the Baseline Apps on Hackintoshes, even a couple that have run the Special App without any trouble. Yours is the only one I've seen give Invalids with the CUDA65 App, which is Why I suggest the CUDA75 App. You can always change the two lines in the app_info.xml back to the CUDA65 App if it isn't any better. It's very simple to copy and paste the App Name in the Two lines that list it. That's All that's required, replace one App name with the other in two lines.

In any case, I would suggest you update the CUDA driver to the latest available and see if the CUDA65 App works any better on your particular Hackintosh. There are other Hackintoshes running even the CUDA80 driver without any trouble.
Oh, just in case, You Must Quit any Apps using the CUDA Driver Before attempting to Update the Driver.