Maxed Out 5,1 cMP Reliability?

[Grumply]

Hi Guys,

After a great deal of pondering, consternation and concern, I decided (thanks to still needing Prores and FCPX) that I couldn't yet leave the Mac platform for the increasingly more appealing world of PCs.

I currently have a maxed-out late 2012 iMac, but it just can't keep up with the video post-production work I'm doing (primarily colour grading), and a recent nightmare project on the iMac cost me so much time, I simply could not afford to carry-on as things were.

I’ve never been a fan of the Trashcan's layout (adding endless PCIe and HDD enclosures and covering my desk in extra cables holds little appeal) and rendering issues with the nMP's D700 cards in Davinci Resolve made it a non-starter for me anyway. So instead I've been pushed down the path of investing $9,000(!) of my hard-earned into an 8-year-old cMP (4,1 flashed to 5,1).

Frankly I find this whole situation rather ludicrous. But thanks to still needing Prores and FCPX, it seemed the best course left to me.

The config of this system is:

- 12-core 3.46Ghz
- 64GB RAM
- Flashed 12GB Titan X (Maxwell)
- 240GB Samsung 850 Evo SSD (Boot Drive, in optical bay)
- 4x 6TB Seagate Ironwolf HDDs (in an internal RAID 5 via Softraid)
- 2x 1.92TB Samsung Enterprise SSDs in RAID 0 on a Sonnet Tempo Pro Plus Card (scratch drive)
- Blackmagic Decklink Mini Monitor 4k I/O Card
- Sonnet Allegro Pro 4-port USB3 Card
- Asus PA329Q 4k Monitor

I’ve never gone quite so all-out with a machine before, but it seemed the only way to make an 8-year-old computer as future-proof as possible.

Now I sourced the base machine from a gent in the UK who refurbishes cMPs for a living:

- 12-core 3.46Ghz
- 64GB RAM
- Flashed 12GB Titan X (Maxwell)
- 240GB Samsung 850 Evo SSD (Boot Drive, in optical bay)
- Sonnet Allegro Pro 4-port USB3 Card

And added the other parts as I got them. However, from the get go, I’ve had serious instabilities with the system. The computer would randomly shut off without warning (and often not even under load), and as I added the other components, this instability appeared to grow worse, along with boot times, which grew out to 45 seconds before the Apple chime would sound - after which the machine would boot very rapidly (10 seconds or so) from the SSD, but the looooong wait before the booting would actual commence became infuriating pretty quickly. Shut down times also grew very long.

Now my first thought was that this might be something to do with the Titan X getting enough power (as the setup from the shop was just taking power from the two mini 6-pin ports on the motherboard). I did some reading on this and bought a few adapters and cables so that I could combine the two mini 6-pins into a single 8-pin connector, and I unplugged the DVD drive in the optical bay and stole its power to feed the 6-pin port on the Titan. All up, this setup should provide 279w to the GPU, which should be enough to keep it running even under full load… but the shutdowns persisted without interruption.

Some of the time, on restart, I got no system reports at all; and the rest of the time, it appears to have been kernel panics that caused the machine to sh*t itself and switch off.

At this stage I threw up my hands and sent the machine back to the UK for a replacement (which will hopefully make its way back down to me soonish), but I was wondering what people’s experiences have been with pushing the cMP so far, with such powerful components? Are instabilities common when the architecture is maxed out? Is it simply more than the base hardware was ever designed to deal with?

If I can get the system running stably, I’m sure I’ll be quite happy with the performance (which is still extremely powerful in spite of the system’s age). But if I can’t get that stability, then it’s no good to me at all. The whole point of investing so much into this machine was to get away from the instabilities and crashes when pushing the iMac so much further than it wants to go.

I’d really appreciate hearing people’s experiences with maxing out the cMPs, and whether I should be worried? Or whether, properly setup, the system can be as bulletproof as my old cMPs 1,1 was?

Cheers,

Mark

 

[Theophany]

$9,000 on a nearly decade-old computer is insanity. There's no other word for it really. Especially considering future MacOS support will probably be dropped this year or next.

 

[orph]

eeek the price

if you spent that much might be worth trying to get support from the guy who sold it.
(for that much you deserve some help)

may be the GPU, if it's a power thing an external power supply will do it.
may be a pci card, pull them one by one till you get no crash.
may be something else can you post temps?
(think this app has a demo http://www.bresink.com/osx/HardwareMonitor.html)
might also be a software problem >.< which may be a pain to track down.

where did you get the GPU from? macvideocards ?
im not shore on how you powered your gpu, the dvd drive plug is not for that kind of load, might be better with two 6pin-to-8pin but not shore.
if it is a flashed card seek help from where you got it.

id have gotten an imac

 

[Grumply]

Insanity? It does kinda feel that way

To be fair, the box itself (with GPU and USB3 card) was only $4,000 of that (and over $1000 of that is just the GPU), and the rest of the parts would port over to another computer directly if need be. So it's not really as bad as it is sounds (and cost-wise, no combination of buying/assembling components from assorted places could have worked out cheaper - which I know because I spreadsheeted them all!). So, on-the-whole it's not an outrageous deal (as far as getting an 8-year-old computer for thousands of dollars can be 'not an outrageous deal'), but with that much dosh on the line, the thing DOES need to work, and work well.

My only consolation with the possible software support becoming an issue, is that Davinci Resolve (my primary software) is so mature as a piece of software now, that if I have to lock everything at the level it's at today, I'll probably be able to keep on going with it for a good while yet for work.

The GPU is a macvideocards card (according to the system report).

Temps never got out of hand (according to iStat Pro nothing has made it much over 60 degrees during any of the shut offs I've had so far).

 

[orph]

did you buy it from macvideocards?
if so contact him and ask for help, he provides support.

something like OctaneBench can be used to stress the cpu (do not use furmark ever even on good GPU's it's bad and most have been gimped by nvida/ati to run at low speed on furmark)
https://render.otoy.com/octanebench/

LuxMark is also worth a goo and has a "stress gpu mode" that will keep the load on, if you have hardware monitor you can watch power usage.
http://www.luxrender.net/wiki/LuxMark#Binaries

if you crash during one of the two then it's a good chance it's the GPU using to much power.

im relay not happy with how you plunged in your gpu, sounds way to unbalance.
the dvd power plug id gess is not designed to give 65-74w of power which is what some GPU's can pull. (i dont know how much titans use)

the price is not good but ok if it provides professional support with the system.

edit if you can post a screen shot it may help of temps/power use

 

[macuser453787]

Hi Guys,

After a great deal of pondering, consternation and concern, I decided (thanks to still needing Prores and FCPX) that I couldn't yet leave the Mac platform for the increasingly more appealing world of PCs.

I currently have a maxed-out late 2012 iMac, but it just can't keep up with the video post-production work I'm doing (primarily colour grading), and a recent nightmare project on the iMac cost me so much time, I simply could not afford to carry-on as things were.

I’ve never been a fan of the Trashcan's layout (adding endless PCIe and HDD enclosures and covering my desk in extra cables holds little appeal) and rendering issues with the nMP's D700 cards in Davinci Resolve made it a non-starter for me anyway. So instead I've been pushed down the path of investing $9,000(!) of my hard-earned into an 8-year-old cMP (4,1 flashed to 5,1).

Frankly I find this whole situation rather ludicrous. But thanks to still needing Prores and FCPX, it seemed the best course left to me.

The config of this system is:

- 12-core 3.46Ghz
- 64GB RAM
- Flashed 12GB Titan X (Maxwell)
- 240GB Samsung 850 Evo SSD (Boot Drive, in optical bay)
- 4x 6TB Seagate Ironwolf HDDs (in an internal RAID 5 via Softraid)
- 2x 1.92TB Samsung Enterprise SSDs in RAID 0 on a Sonnet Tempo Pro Plus Card (scratch drive)
- Blackmagic Decklink Mini Monitor 4k I/O Card
- Sonnet Allegro Pro 4-port USB3 Card
- Asus PA329Q 4k Monitor

I’ve never gone quite so all-out with a machine before, but it seemed the only way to make an 8-year-old computer as future-proof as possible.

Now I sourced the base machine from a gent in the UK who refurbishes cMPs for a living:

- 12-core 3.46Ghz
- 64GB RAM
- Flashed 12GB Titan X (Maxwell)
- 240GB Samsung 850 Evo SSD (Boot Drive, in optical bay)
- Sonnet Allegro Pro 4-port USB3 Card

And added the other parts as I got them. However, from the get go, I’ve had serious instabilities with the system. The computer would randomly shut off without warning (and often not even under load), and as I added the other components, this instability appeared to grow worse, along with boot times, which grew out to 45 seconds before the Apple chime would sound - after which the machine would boot very rapidly (10 seconds or so) from the SSD, but the looooong wait before the booting would actual commence became infuriating pretty quickly. Shut down times also grew very long.

Now my first thought was that this might be something to do with the Titan X getting enough power (as the setup from the shop was just taking power from the two mini 6-pin ports on the motherboard). I did some reading on this and bought a few adapters and cables so that I could combine the two mini 6-pins into a single 8-pin connector, and I unplugged the DVD drive in the optical bay and stole its power to feed the 6-pin port on the Titan. All up, this setup should provide 279w to the GPU, which should be enough to keep it running even under full load… but the shutdowns persisted without interruption.

Some of the time, on restart, I got no system reports at all; and the rest of the time, it appears to have been kernel panics that caused the machine to sh*t itself and switch off.

At this stage I threw up my hands and sent the machine back to the UK for a replacement (which will hopefully make its way back down to me soonish), but I was wondering what people’s experiences have been with pushing the cMP so far, with such powerful components? Are instabilities common when the architecture is maxed out? Is it simply more than the base hardware was ever designed to deal with?

If I can get the system running stably, I’m sure I’ll be quite happy with the performance (which is still extremely powerful in spite of the system’s age). But if I can’t get that stability, then it’s no good to me at all. The whole point of investing so much into this machine was to get away from the instabilities and crashes when pushing the iMac so much further than it wants to go.

I’d really appreciate hearing people’s experiences with maxing out the cMPs, and whether I should be worried? Or whether, properly setup, the system can be as bulletproof as my old cMPs 1,1 was?

Cheers,

Mark

From a troubleshooting standpoint, the issue sounds like it could be one of two things:

-- Something about the base config components, or

-- Something about the cMP itself.


If you get the same base config again and the issues do not repeat themselves, it was likely something about the first cMP itself.

If you get the same base config again and the issues do repeat themselves, it is likely something about the components. In that case, it might be good to start with removing the GPU and the USB3 card one at a time and see if the issue comes back.

 

[kalex]

did you buy it from macvideocards?
if so contact him and ask for help, he provides support.

Is he still in business? he has been banned from here and no inventory refresh on his site. twitter also quite
[doublepost=1487945116][/doublepost]Op - I have upgraded 2009 model and never had stability issues. Actually I did when i tried to upgrade the CPU. During the process I accidently snipped of one of the northbridge heatsync's pins and after 4 months my Mac Pro died. It wouldn't boot or anything. Swapping the tray out fixed the issue. Mine is not as full as yours but I do have after market ram, flashed gtx 680, usb 3 card and pcie ssd boot drive. So far no issues. Upgrading few parts made a big difference and my next step is to get new cpu and video card

 

[Grumply]

did you buy it from macvideocards?
if so contact him and ask for help, he provides support.

something like OctaneBench can be used to stress the cpu (do not use furmark ever even on good GPU's it's bad and most have been gimped by nvida/ati to run at low speed on furmark)
https://render.otoy.com/octanebench/

LuxMark is also worth a goo and has a "stress gpu mode" that will keep the load on, if you have hardware monitor you can watch power usage.
http://www.luxrender.net/wiki/LuxMark#Binaries

if you crash during one of the two then it's a good chance it's the GPU using to much power.

im relay not happy with how you plunged in your gpu, sounds way to unbalance.
the dvd power plug id gess is not designed to give 65-74w of power which is what some GPU's can pull. (i dont know how much titans use)

the price is not good but ok if it provides professional support with the system.

edit if you can post a screen shot it may help of temps/power use

Thanks, the benchmarks sound like a good option for stress testing the machine (certainly should be more consistent than just running renders in Davinci). I'll give them a try when the replacement arrives.

As I understand it, the Titan would be getting 75w from the PCIe port, 150w from the dual mini 6-pins feeding into the 8-pin port on the GPU, and another 54w from the SATA power sourced from the optical bay. It would be nice to get a full 75w feeding into the 6-pin port, but the overall supply should theoretically be enough (and I'd assume the card would draw power from the PCIe port and the 8-pin port in priority over the 6-pin port). At any rate, if the gpu stress test seems to crash things, I'll just move to an external PSU for the Titan, to eliminate that as a possible root cause.

From a troubleshooting standpoint, the issue sounds like it could be one of two things:

-- Something about the base config components, or

-- Something about the cMP itself.


If you get the same base config again and the issues do not repeat themselves, it was likely something about the first cMP itself.

If you get the same base config again and the issues do repeat themselves, it is likely something about the components. In that case, it might be good to start with removing the GPU and the USB3 card one at a time and see if the issue comes back.

Thanks, I'll certainly try to stress test the system with each additional component as I add them in, to see if anything in particular causes the shutdowns (I'm hopeful the replacement will just work, and the nightmare will be over). I was getting very strange little things happening like the USB ports (either the built-in USB2s or the added USB3s) becoming unresponsive after a shutdown, and having to restart the computer to get them running again. And sometimes the computer would start up and the machine wouldn't recognise the presence of the HDD in my first SATA drive bay. And all of these little oddities would happen either individually, or together, or not at all. It all appeared to be quite random. Sometime on a reboot, the login screen would come up, but the image on my monitor was complete corrupted and would look like some kind of psychedelic kaleidoscope until I unplugged the monitor and then plugged it back in.
[doublepost=1487945312][/doublepost]

Is he still in business? he has been banned from here and no inventory refresh on his site. twitter also quite
[doublepost=1487945116][/doublepost]Op - I have upgraded 2009 model and never had stability issues. Actually I did when i tried to upgrade the CPU. During the process I accidently snipped of one of the northbridge heatsync's pins and after 4 months my Mac Pro died. It wouldn't boot or anything. Swapping the tray out fixed the issue. Mine is not as full as yours but I do have after market ram, flashed gtx 680, usb 3 card and pcie ssd boot drive. So far no issues. Upgrading few parts made a big difference and my next step is to get new cpu and video card

Thanks Kalex, that's reassuring.

I certainly read plenty of happy user reports of maxed out cMPs when I started investigating whether I should consider one. And my old 1,1 MP had been flawless for the six years I ran it, so I felt like it would be straight forward.

It's a lot of high-end hardware to shove in an old machine though, so I do want to hear whether anyone's run into similar issues before.

 

[h9826790]

Interesting choice, you pick a very expensive graphic card and USB 3.0 card which doesn't looks can be utilised.

e.g. AMD cards are well known work better in FCPX. And I doubt for normal small scale video production, why do you need such an expensive USB 3.0 card. I just can't imagine that you will continuously fully utilise all 4 ports at the same time. If yes, the iMac should be way too limiting to you long long time ago.

IMO, even $4000 for that config is way too expensive. But that's mainly true in the US market or in my home town. May be that's a good price in your place.

Back to the topic, system info said this is a MVC TitanX, not necessary mean it's the truth. Quite a few seller out there just dump the ROM from the real MVC card and then flash it to other cards. They have no idea what have to be done and how to test, fine tune, etc. In some case, the card itself can cause lots of stability (not because of power problem, but simply because of poorly flashed).

Since you have another Mac, it's very easy to know if the instability is GPU related. Simply setup remote control, remove the GPU, boot the Mac Pro headless, and remote control your Mac Pro by the iMac. Of course, I am not asking you to work in this way. However, if the Mac Pro can function normally, or even survive under stress test for days. Then, it's highly likely the instability is GPU related. Also, the RAMs are also the suspect if any trouble happened randomly from time to time.

Anyway, the cMP is a very stable machine. I bought my 4,1 directly from Apple when it was newly released. And it works flawlessly at this moment with the upgrade stated in my signature. It's a very nice FCPX machine, and I use it hard 24/7. No self reboot, no crash, no instability. Just keep running 24/7, that's it. And I know there are few guys here running a maxed out cMP (similar to yours) without any trouble as well.

Also, your assumption is wrong. The TitanX may draw UP TO 75W from the slot, 150W from the 8pin, and 75W from the 6pin (without software OC, or any mod to the ROM). The power draw is base on the card, not your supply. If the graphic card decide to draw 75W from the 6pin, it will draw 75W from the SATA port in your case, which may shutdown the Mac, or kill the port, etc.

Anyway, since you have iStat, can you post some screen capture of the temperatures? at least one during idle, and another one under stress. Which may help us to know more if something look abnormal there.

 

[macuser453787]

Reviewing MVC's web page for your GPU, the MVC-flashed Titan X 12GB Maxwell runs on internal power.

So if it is an MVC card then it would seem that the issue could have been with the cMP itself.

Here is an interesting post where the user was experiencing random shutdowns. In their case it was due to a bad RAM riser card, but the suggestions may be worth noting/checking if the problem persists: possible bad RAM module, possible power supply issue...

FWIW I've been running a highly upgraded (though not quite as extensively as yours) 4,1-->5,1 cMP for some years now and it's been rock solid.

One other thing that comes to mind is that I had a similar boot delay (about 1 minute between pressing the power button and the Apple symbol showing on screen) when I was using a Sonnet Tempo Pro Plus Card as my boot drive, with 2 Samsung 840 Pro 256GB's in MacOS RAID 0. I know you're using yours as a scratch disk, but your description of the increased boot time reminds me of that. If that issue persists in your replacement cMP, you might try taking the Sonnet card out and see if the boot time improves.

 

[Grumply]

Interesting choice, you pick a very expensive graphic card and USB 3.0 card which doesn't looks can be utilised.

e.g. AMD cards are well known work better in FCPX. And I doubt for normal small scale video production, why do you need such an expensive USB 3.0 card. I just can't imagine that you will continuously fully utilise all 4 ports at the same time. If yes, the iMac should be way too limiting to you long long time ago.

IMO, even $4000 for that config is way too expensive. But that's mainly true in the US market or in my home town. May be that's a good price in your place.

The Titan isn't really for FCPX, I don't need lots of grunt for editing, it's for colour grading in Davinci Resolve (where I need as much GPU RAM and processing power as possible for pushing around 4k and 6k raw files, and CUDA cores have a huge impact on performance).

And the USB3 card offers individual lanes for each port, so when I have to plug a bunch of media cards into the machine, I have full-speed access to all of them.

The $4000 is in Australian dollars (and includes a couple of hundred in shipping costs) that's around $3,000 USD. I'd love to hear where you can source the same thing cheaper - because I scoured the internet for weeks to find the cheapest route to this particular build (including getting the components separately and doing the build myself), and this was the most affordable I could find!

Back to the topic, system info said this is a MVC TitanX, not necessary mean it's the truth. Quite a few seller out there just dump the ROM from the real MVC card and then flash it to other cards. They have no idea what have to be done and how to test, fine tune, etc. In some case, the card itself can cause lots of stability (not because of power problem, but simply because of poorly flashed).

Since you have another Mac, it's very easy to know if the instability is GPU related. Simply setup remote control, remove the GPU, boot the Mac Pro headless, and remote control your Mac Pro by the iMac. Of course, I am not asking you to work in this way. However, if the Mac Pro can function normally, or even survive under stress test for days. Then, it's highly likely the instability is GPU related. Also, the RAMs are also the suspect if any trouble happened randomly from time to time.

Anyway, the cMP is a very stable machine. I bought my 4,1 directly from Apple when it was newly released. And it works flawlessly at this moment with the upgrade stated in my signature. It's a very nice FCPX machine, and I use it hard 24/7. No self reboot, no crash, no instability. Just keep running 24/7, that's it. And I know there are few guys here running a maxed out cMP (similar to yours) without any trouble as well.

Also, your assumption is wrong. The TitanX may draw UP TO 75W from the slot, 150W from the 8pin, and 75W from the 6pin (without software OC, or any mod to the ROM). The power draw is base on the card, not your supply. If the graphic card decide to draw 75W from the 6pin, it will draw 75W from the SATA port in your case, which may shutdown the Mac, or kill the port, etc.

Anyway, since you have iStat, can you post some screen capture of the temperatures? at least one during idle, and another one under stress. Which may help us to know more if something look abnormal there.

That's all really helpful, thank you. I think if the GPU stability seems to be an issue, I'll just move to an external PSU to ensure that the card is properly taken care of.

My iStat screengrabs are gone with the computer, but the hottest thing I saw was one of the CPUs at 64 degrees celsius, and the GPU at around 50. So nothing seemed too extreme.

Reviewing MVC's web page for your GPU, the MVC-flashed Titan X 12GB Maxwell runs on internal power.

So if it is an MVC card then it would seem that the issue could have been with the cMP itself.

Here is an interesting post where the user was experiencing random shutdowns. In their case it was due to a bad RAM riser card, but the suggestions may be worth noting/checking if the problem persists: possible bad RAM module, possible power supply issue...

FWIW I've been running a highly upgraded (though not quite as extensively as yours) 4,1-->5,1 cMP for some years now and it's been rock solid.

One other thing that comes to mind is that I had the same boot delay when I was using a Sonnet Tempo Pro Plus Card as my boot drive, with 2 Samsung 840 Pro 256GB's in MacOS RAID 0. I know you're using yours as a scratch disk, but your description of the increased boot time reminds me of that. If that issue persists in your replacement cMP, you might try taking the Sonnet card out and see if the boot time improves.

Thanks. I'll definitely compare the start up times with and without the Sonnet Tempo Pro (strange it should cause the delays even when it's not being used for the boot drive).

I've been through that thread (and all the others I could find) already, and I also kinda suspect it might be the RAM that's playing up. That seemed to be the most common fix to random shutdowns from the advice I could find.

 

[h9826790]

The Titan isn't really for FCPX, I don't need lots of grunt for editing, it's for colour grading in Davinci Resolve (where I need as much GPU RAM and processing power as possible for pushing around 4k and 6k raw files, and CUDA cores have a huge impact on performance).

And the USB3 card offers individual lanes for each port, so when I have to plug a bunch of media cards into the machine, I have full-speed access to all of them.

The $4000 is in Australian dollars (and includes a couple of hundred in shipping costs) that's around $3,000 USD. I'd love to hear where you can source the same thing cheaper - because I scoured the internet for weeks to find the cheapest route to this particular build (including getting the components separately and doing the build myself), and this was the most affordable I could find!




That's all really helpful, thank you. I think if the GPU stability seems to be an issue, I'll just move to an external PSU to ensure that the card is properly taken care of.

My iStat screengrabs are gone with the computer, but the hottest thing I saw was one of the CPUs at 64 degrees celsius, and the GPU at around 50. So nothing seemed too extreme.



Thanks. I'll definitely compare the start up times with and without the Sonnet Tempo Pro (strange it should cause the delays even when it's not being used for the boot drive).

I've been through that thread (and all the others I could find) already, and I also kinda suspect it might be the RAM that's playing up. That seemed to be the most common fix to random shutdowns from the advice I could find.

I know the advantage of that USB 3.0 card, and that's why I wonder how often you will fully utilise all 4 ports at the same time (about 2000MB/s in total). You simply don't have such high performance storage to provide data source at that speed. For most media cards, even using all 4 ports at the same time, still can't quite saturate a $20 KT4004's bandwidth.

Anyway, it's just personal choice. However, if you choose that for future proof, IMO, it's not a good choice. The main "future proof" on the cMP is coming from the upgradability, not the current parts' function / performance. If a KT4004 is good enough for your current usage now, pick that one, keep the extra money. And once the world move to USB-C, get a new USB-C card. That's how to future proof work on the cMP. IMO, the Allegro Pro is an excellent performing card now, but it won't give you any future proof, because it won't compatible to the future standard (e.g. USB-C).

 

[Grumply]

I know the advantage of that USB 3.0 card, and that's why I wonder how often you will fully utilise all 4 ports at the same time (about 2000MB/s in total). You simply don't have such high performance storage to provide data source at that speed. For most media cards, even using all 4 ports at the same time, still can't quite saturate a $20 KT4004's bandwidth.

Anyway, it's just personal choice. However, if you choose that for future proof, IMO, it's not a good choice. The main "future proof" on the cMP is coming from the upgradability, not the current parts' function / performance. If a KT4004 is good enough for your current usage now, pick that one, keep the extra money. And once the world move to USB-C, get a new USB-C card. That's how to future proof work on the cMP. IMO, the Allegro Pro is an excellent performing card now, but it won't give you any future proof, because it won't compatible to the future standard (e.g. USB-C).

Sure. But the card is really only there for offloading footage from camera cards, and for doing so quicker than I can via the USB2 ports. And since the readers for all of those cards are USB3, it's a necessity for the time being.

But it's also the single cheapest component of the entire build, so it's the one part I'm NOT worried about!

 

[h9826790]

Sure. But the card is really only there for offloading footage from camera cards, and for doing so quicker than I can via the USB2 ports. And since the readers for all of those cards are USB3, it's a necessity for the time being.

But it's also the single cheapest component of the entire build, so it's the one part I'm NOT worried about!

May be you misunderstand me a bit. The $20 KT4004 is also a USB 3.0 card, which is good enough to load 4 media card at the same time. Just a lot cheaper.

Anyway, as you said, it's not your concern. And I don't think it's the source of the instability. So, it's actually quite irrelevant in this thread

 

[orph]

i think the MVC titan runs on the dual 6 pins power and has it's bios altered to lower power use (and maybe a hardware mod?)
idk if he's still around, he used to get banned all the time, then came back.

on the product page it only lists two power cables to power it, did your seller give instructions with the card
http://www.macvidcards.com/store/p44/Nvidia_GTX_Titan_X_12_GB_(Maxwell).html

it may also be a simple software problem

 

[macuser453787]

I'll definitely compare the start up times with and without the Sonnet Tempo Pro (strange it should cause the delays even when it's not being used for the boot drive).

I agree, and that may not even be the cause of the longer boot time, but may be worth checking anyway if the issue comes back in your replacement cMP.
[doublepost=1487952060][/doublepost]

I've been through that thread (and all the others I could find) already, and I also kinda suspect it might be the RAM that's playing up. That seemed to be the most common fix to random shutdowns from the advice I could find.

Just curious: did you try testing the RAM modules, as in booting/operating with only one module installed at a time?

 

[ActionableMango]

Are instabilities common when the architecture is maxed out?

No, this is not normal. You mention kernel panics. What do they say? This is potentially critically important information and it could point right at the problem.

 

[itdk92]

Actually today I have experienced an issue I did not have in a week.

FCPX footage is super stuttering/glitchy when using 2 x Titan X

1 GPU (the one or the other) = no stutter

Footage from the HDD/SSD, youtube exc = no stutter

web drivers have been reinstalled, disk checked for errors


The fun part -> with ONE gpu, FCPX does NOT stutter


Any ideas?


****
to better describe this stutter :
it's the playback that's glitching, interface exc run perfectly smooth

 

[Blair Paulsen]

Put in an aux PSU. That Maxwell TitanX/12GB GPU might be able to run on stock internal power, but IMO it's a poor configuration choice. I chose to run a 4GB GTX 980 because it's TDP fits into the stock design parameters. I also am of the opinion that the totality of the system architecture - buses/RAM clock/caches/etc make the higher cost and complexity (putting in an aux PSU can be a PITA) of the big Titan weak ROI. YMMV.

 

[itdk92]

Put in an aux PSU. That Maxwell TitanX/12GB GPU might be able to run on stock internal power, but IMO it's a poor configuration choice. I chose to run a 4GB GTX 980 because it's TDP fits into the stock design parameters. I also am of the opinion that the totality of the system architecture - buses/RAM clock/caches/etc make the higher cost and complexity (putting in an aux PSU can be a PITA) of the big Titan weak ROI. YMMV.

There is no need for auxiliary PSU for either 1 or 2 Titan Xs, especially after pixlas mod, which makes it easier to have the PSU as the direct power source.

So that is not the problem

Also, we got them very cheap, so it was a no-brainer.

The funky part is that we now have "mixed" 1 Titan Xs + GTX980Ti and there is absolutely no glitch!

Pretty random..

***
Also, pretty random, but 2 x GTX980Ti also caused the same glitch in FCPX... might there be a problem with dual Nvidia gpu setup where the gpus are the same model?

Titan X + GTX980Ti seems to work flawlessly

 

[Grumply]

Hi guys,

Just to update and close-off this thread. I sent it back to the seller and paid a bit extra to switch out the flashed 4,1 chassis for a genuine 2010 5,1 chassis. Received the replacement back a week ago, and it's been rock-solid since. Hasn't had so much as a hiccup under load or running heavy benchmarks (even stuffed full of all of the extra components).

I guess it must have been something to do with the flashed bios on the 4,1 chassis.

So to anyone else pondering a similar path to a 'powerful' mac, my advice would be avoid the flashed 4,1s and stick to the proper 2010-2012 5,1 models.

Cheers,

Mark

 

[LightBulbFun]

Just to throw my 2 cent in here You mention you had a flashed 4,1 shipped from the UK to Auz, unless he used LidLess CPUs he would of had to of done the Fun Rigmarole of getting the screws just the right tightness with lidded CPUs and id not be surprised if A he did not do it right in the first place and or B during shipping, since the heatsinks are never seated properly with lidded CPUs in a lidless tray, the screws or the heatsinks them selfs could of been knocked lose during shipping and dislodged the CPUs. I have seen a few reports of flashed 4,1 Mac Pros shutting down under load especially due to that, which was solved by tightening down the heatsinks that little bit more

another cause could of been the plastic rivets/push pins on the PCH heatsink breaking lose during shipping or just over time and when the heatsink pops lose the PCH overheats and the MP shuts down thats an issue with the 4,1 and 5,1 that can crop up

 

[AndreeOnline]

Using a flashed 4.1 --> 5.1 myself that does regular rendering both on-the-fly in FCPX and DaVinci as well as 3D in Arnold and Maxwell and it doesn't break a sweat even after many hours of full load.

The ageing cMP does have a few issues, but freaking out under pressure isn't one of them.

 

[itdk92]

Hi guys,

Just to update and close-off this thread. I sent it back to the seller and paid a bit extra to switch out the flashed 4,1 chassis for a genuine 2010 5,1 chassis. Received the replacement back a week ago, and it's been rock-solid since. Hasn't had so much as a hiccup under load or running heavy benchmarks (even stuffed full of all of the extra components).

I guess it must have been something to do with the flashed bios on the 4,1 chassis.

So to anyone else pondering a similar path to a 'powerful' mac, my advice would be avoid the flashed 4,1s and stick to the proper 2010-2012 5,1 models.

Cheers,

Mark

This is just not true.

Technical spec between the models are equal.

Sounds like the other machine you own had a problem with a component.

 

[h9826790]

Agree that 4,1 may be faulty, nothing to do with the 5,1 firmware flashing.

There is no real difference in spec, performance, abilities between 4,1 and 5,1.

There are plenty of members here (including myself) have flashed 5,1 and able to run them 24/7 without any trouble.