Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Status
The first post of this thread is a WikiPost and can be edited by anyone with the appropiate permissions. Your edits will be public.
Thanks, I will check it out. BTW I edited the post above after learning that typing fast made 3840x1080p unworkable over DP. Sigh. I did not mind 3840x1080 much at all but would like to have a card capable of powering the display at native, if possible.
I have the same GPU and run the card just fine, even played video games on it under macOS at that resolution. It might be some OS issues. I hope the app solves it for you.
 
  • Like
Reactions: NC12
UPDATE: After Pixlas mod, I am back to getting frequent uncommanded video death when working in Preview. Troubleshooting with ChatGPT has led to recommendation that I buy a new video card, as the RX580 in here is 5-6 years old. What should I get to run the 5120x1440 reliably with document heavy workload. Crashes always happen when opening, scrolling, resizing or moving Quicklook or Preview windows. Running Mojave and am inclined to stick with it on this machine.
 
UPDATE: After Pixlas mod, I am back to getting frequent uncommanded video death when working in Preview. Troubleshooting with ChatGPT has led to recommendation that I buy a new video card, as the RX580 in here is 5-6 years old. What should I get to run the 5120x1440 reliably with document heavy workload. Crashes always happen when opening, scrolling, resizing or moving Quicklook or Preview windows. Running Mojave and am inclined to stick with it on this machine.
For most people the Pixlas mod is not necessary. What is the TDP of your card?

The crashing seems like a dying gpu or some sort of issue because of the mod. Did you try putting the oem gpu back in to test?

ChatGPT’s suggestion is bs honestly. A 580 is more than enough for preview or any modern office work at that resolution. My guess would be some sort of power issue or a defective gpu
 
Thanks, that's helpful because the blackscreens (but Mac stays on; and but can't fix with SSH in and sudo killall Windowserver) were occurring with 4k 32" panel before, moving to 49" 5120x1440 only accelerated), and Pixlas mod didn't seem to help. Thinking of how the issue has accelerated even after I moved to this monitor (tonight I had a crash from switching tabs in preview which never happened before) convinces me it's worth trying a fresh GPU.

I don't know what TDP means? I am now shopping for a new RX580.

Do you know if these single slot RX580s will work?
 
As an Amazon Associate, MacRumors earns a commission from qualifying purchases made through links in this post.
Is your Mojave the latest version 18G9323 ?
Try on a fresh clean install of Mojave and see if the issues follow. 5-6 yrs old is not old if the thermal paste is still wet.
Could be an outdated Quicklook binary, i had to update mine to be able to view .webp files and its great now.
 
I don't know what TDP means? I am now shopping for a new RX580.

Thermal Design Power. Basically, the maximum amount of heat a component can produce / electrical power it can draw. You should be able to power a regular RX 580 from the two mini 6-pin connectors near the GPU on the motherboard (+75W from the slot itself). The recommended model back in the day was the Sapphire Pulse version. I used one for a few years in an unmodified 5,1 and never had any issues.

Personally, I would rather put the money in an AS Mac or PC, though. What software is keeping you on Mojave?
 
  • Like
Reactions: NC12
Thermal Design Power. Basically, the maximum amount of heat a component can produce / electrical power it can draw. You should be able to power a regular RX 580 from the two mini 6-pin connectors near the GPU on the motherboard (+75W from the slot itself). The recommended model back in the day was the Sapphire Pulse version. I used one for a few years in an unmodified 5,1 and never had any issues.

Personally, I would rather put the money in an AS Mac or PC, though. What software is keeping you on Mojave?
Mail indexing + Houdahspot
 
Mail indexing + Houdahspot

According to Houdahspot's website, Apple Mail search is no longer available since macOS Sonoma. That would still allow you to run it on Ventura + an M1 Mac. Presumably this app benefits mostly from single thread CPU speed and SSD bandwidth? Even a mini would be better equipped in that regard than a 5,1.

Is this a one-and-done task, or an ongoing one? If the latter, is there no alternative app that runs on a recent OS?
 
According to Houdahspot's website, Apple Mail search is no longer available since macOS Sonoma. That would still allow you to run it on Ventura + an M1 Mac. Presumably this app benefits mostly from single thread CPU speed and SSD bandwidth? Even a mini would be better equipped in that regard than a 5,1.

Is this a one-and-done task, or an ongoing one? If the latter, is there no alternative app that runs on a recent OS?
Not if you have a quad m.2 16x pcie card and RAID 0 them like i do, you'll easily double the speeds of an M1 Mac.
 
Not if you have a quad m.2 16x pcie card and RAID 0 them like i do, you'll easily double the speeds of an M1 Mac.

OK, but does he have a quad M.2 card in RAID 0 like you do? And even if he does, is the 5,1 actually faster than an M1 mini, given the latter has ~triple the ST performance? And even then, does it make a material difference, especially given the massive disparity in size and energy consumption?
 
Not if you have a quad m.2 16x pcie card and RAID 0 them like i do, you'll easily double the speeds of an M1 Mac.
I'm running off an 8TB WD Black SN850X NVME off a PCI controller that claims to provide 8.0 GT/s off of slot 2. The storage is important to me for ediscovery tasks. Mojave is pretty snappy. I don't have to frequently search email using Houdahspot, but when I need it, I need it. I'm pretty sure the machine is maxed at Mojave unless I install OpenCore which I have been reluctant to do.

The Quicklook/Preview crashing Windowserver issue seems to really be getting out of control. The same pdf that caused a crash last night when I was trying to resize the window (flicker, flicker, video output dead and audio died too this time) caused a crash this morning when I just tried to open it. The pdf is a court exhibit that contains photos of my client. I can do anything else like move browser and video windows around with no problem. It's only PDFs with Preview and Quicklook that sometimes causes the crashes. Some PDFs like short black and white documents seem to cause no problems. Many do, especially if they contain pictures.
 
I'm running off an 8TB WD Black SN850X NVME off a PCI controller that claims to provide 8.0 GT/s off of slot 2.

Are you using a card with a PLX switch, that can offer the full bandwidth of a PCIe 2.0 16x slot to a single PCIe 4.0 4x M.2 SSD?

Nice SSD by the way; I've got two in smaller capacities. Hope you bought it a while ago!
 
I'm running off an 8TB WD Black SN850X NVME off a PCI controller that claims to provide 8.0 GT/s off of slot 2. The storage is important to me for ediscovery tasks. Mojave is pretty snappy. I don't have to frequently search email using Houdahspot, but when I need it, I need it. I'm pretty sure the machine is maxed at Mojave unless I install OpenCore which I have been reluctant to do.

The Quicklook/Preview crashing Windowserver issue seems to really be getting out of control. The same pdf that caused a crash last night when I was trying to resize the window (flicker, flicker, video output dead and audio died too this time) caused a crash this morning when I just tried to open it. The pdf is a court exhibit that contains photos of my client. I can do anything else like move browser and video windows around with no problem. It's only PDFs with Preview and Quicklook that sometimes causes the crashes. Some PDFs like short black and white documents seem to cause no problems. Many do, especially if they contain pictures.
And this is with your new graphics card? 8GT/s just means that's a pcie 3.0 card. We don't actually go past 5GT/s.
 
Are you using a card with a PLX switch, that can offer the full bandwidth of a PCIe 2.0 16x slot to a single PCIe 4.0 4x M.2 SSD?

Nice SSD by the way; I've got two in smaller capacities. Hope you bought it a while ago!
3.0, i bought it this year when prices finally dropped low enough.
 
OK, but does he have a quad M.2 card in RAID 0 like you do? And even if he does, is the 5,1 actually faster than an M1 mini, given the latter has ~triple the ST performance? And even then, does it make a material difference, especially given the massive disparity in size and energy consumption?
Ok, but does he have an M1 Mac like you do?
 
And this is with your new graphics card? 8GT/s just means that's a pcie 3.0 card. We don't actually go past 5GT/s.
I'm losing track of what the question is. I have a CmP 5,1 dual processor with 8TB SN850X as the boot drive. Consequently, it is fast for my edsiscovery use cases which include python3 / regex search work on multi-terabyte file sets. Per ChatGPT, I would need a pretty top spec Apple Silicon machine to equal the speed and/or utilize the drive, and have the graphics power to run the 49" 5120x1440 display, much less at HDR10 which the display is capable of but I'm not sure if I am getting now.

I do need a new laptop. And it would be nice to get one that could connect to the SN850X in an enclosure so I could continue to utilize it without a speed bottleneck. I believe it said that Thunderbolt 5 on the new Macbook Pro M5 Pro and M5 Max is theoretically that but that enclosures aren't available yet.

But the bigger issue I need to solve is I want to keep using the new display, but I need to resolve the issue of Preview and Quicklook crashing my machine and requiring a power-off (button) restart all the time. The problem existed before when I was running a 32" 4k panel. But it was rare. With this new display at 5120x1440, it is not rare at all and it seems to be getting more frequent. Court documents are all in PDF. So PDFs are a critical part of my workflow.

I feel strongly that there *should& be a software fix for this. But I can't seem to find it, notwithstanding that there appear to be other reports of this happening to users with larger displays, and that the problem many not be limited to Mojave. https://discussions.apple.com/thread/255031336

But the direction ChatGPT has pushed in is that the crashes are due to Preview and Quicklook putting a special kind of strain on the video card, that while the XFX RX580 I've been using for the past 5-6 years in this machine is enough to power the display at 60 or 120 hz, it's not powerful enough to avoid window server crashing from Preview/Quicklook:
2026-05-25 20:16:34.883026-0700 localhost Preview[461]: (QuartzCore) Created Activity ID: 0x15500c, Description: Load Metal shader
2026-05-25 20:16:34.883052-0700 localhost Preview[461]: (QuartzCore) Created Activity ID: 0x15500d, Description: Load Metal shader
2026-05-25 20:16:34.883078-0700 localhost Preview[461]: (QuartzCore) Created Activity ID: 0x15500e, Description: Compile Metal pipeline
2026-05-25 20:16:34.884217-0700 localhost Preview[461]: (QuartzCore) Created Activity ID: 0x15500f, Description: Load Metal shader
2026-05-25 20:16:34.884228-0700 localhost Preview[461]: (QuartzCore) Created Activity ID: 0x155010, Description: Load Metal shader
2026-05-25 20:16:34.884232-0700 localhost Preview[461]: (QuartzCore) Created Activity ID: 0x155011, Description: Compile Metal pipeline
2026-05-25 20:16:34.884295-0700 localhost Preview[461]: (QuartzCore) Created Activity ID: 0x155012, Description: Load Metal shader
2026-05-25 20:16:34.884300-0700 localhost Preview[461]: (QuartzCore) Created Activity ID: 0x155013, Description: Load Metal shader
2026-05-25 20:16:34.884320-0700 localhost Preview[461]: (QuartzCore) Created Activity ID: 0x155014, Description: Compile Metal pipeline
2026-05-25 20:16:39.056608-0700 localhost kernel[0]: (IOAcceleratorFamily2) void IOAccelFenceMachine::fence_timeout(IOTimerEventSource *): AMDRadeonAccelerator prodding blockFenceInterrupt
2026-05-25 20:16:39.157599-0700 localhost kernel[0]: (IOAcceleratorFamily2) void IOAccelFenceMachine::fence_timeout(IOTimerEventSource *): AMDRadeonAccelerator prodding blockFenceInterrupt
2026-05-25 20:16:39.258599-0700 localhost kernel[0]: (IOAcceleratorFamily2) void IOAccelFenceMachine::fence_timeout(IOTimerEventSource *): AMDRadeonAccelerator prodding blockFenceInterrupt
2026-05-25 20:16:39.359646-0700 localhost kernel[0]: (IOAcceleratorFamily2) void IOAccelFenceMachine::fence_timeout(IOTimerEventSource *): AMDRadeonAccelerator prodding blockFenceInterrupt
2026-05-25 20:16:39.460681-0700 localhost kernel[0]: (IOAcceleratorFamily2) void IOAccelFenceMachine::fence_timeout(IOTimerEventSource *): AMDRadeonAccelerator prodding blockFenceInterrupt
2026-05-25 20:16:39.561666-0700 localhost kernel[0]: (IOAcceleratorFamily2) void IOAccelFenceMachine::fence_timeout(IOTimerEventSource *): AMDRadeonAccelerator prodding blockFenceInterrupt
2026-05-25 20:16:39.662666-0700 localhost kernel[0]: (IOAcceleratorFamily2) void IOAccelFenceMachine::fence_timeout(IOTimerEventSource *): AMDRadeonAccelerator prodding blockFenceInterrupt
2026-05-25 20:16:39.763675-0700 localhost kernel[0]: (IOAcceleratorFamily2) void IOAccelFenceMachine::fence_timeout(IOTimerEventSource *): AMDRadeonAccelerator prodding blockFenceInterrupt
2026-05-25 20:16:39.864239-0700 localhost kernel[0]: (IOAcceleratorFamily2) void IOAccelFenceMachine::fence_timeout(IOTimerEventSource *): AMDRadeonAccelerator prodding blockFenceInterrupt
2026-05-25 20:16:39.964318-0700 localhost kernel[0]: (IOAcceleratorFamily2) void IOAccelFenceMachine::fence_timeout(IOTimerEventSource *): AMDRadeonAccelerator prodding blockFenceInterrupt
2026-05-25 20:16:40.064379-0700 localhost kernel[0]: (IOAcceleratorFamily2) void IOAccelFenceMachine::fence_timeout(IOTimerEventSource *): AMDRadeonAccelerator prodding blockFenceInterrupt
2026-05-25 20:16:40.165384-0700 localhost kernel[0]: (IOAcceleratorFamily2) void IOAccelFenceMachine::fence_timeout(IOTimerEventSource *): AMDRadeonAccelerator prodding blockFenceInterrupt
2026-05-25 20:16:40.266377-0700 localhost kernel[0]: (IOAcceleratorFamily2) void IOAccelFenceMachine::fence_timeout(IOTimerEventSource *): AMDRadeonAccelerator prodding blockFenceInterrupt
2026-05-25 20:16:40.367371-0700 localhost kernel[0]: (IOAcceleratorFamily2) void IOAccelFenceMachine::fence_timeout(IOTimerEventSource *): AMDRadeonAccelerator prodding blockFenceInterrupt
2026-05-25 20:16:40.468407-0700 localhost kernel[0]: (IOAcceleratorFamily2) void IOAccelFenceMachine::fence_timeout(IOTimerEventSource *): AMDRadeonAccelerator prodding blockFenceInterrupt
2026-05-25 20:16:40.569439-0700 localhost kernel[0]: (IOAcceleratorFamily2) void IOAccelFenceMachine::fence_timeout(IOTimerEventSource *): AMDRadeonAccelerator prodding blockFenceInterrupt
(to infinity)
 
3.0, i bought it this year when prices finally dropped low enough.

I was referring to what @AppleDrank said. Basically - does he have a RAID card like yours, or is he just using a passive adapter? If the latter, he's only getting ~1500MB/s anyway, i.e. 4x PCIe 2.0. Even a cheap USB 3.2 gen 2 enclosure will give 1000MB/s, and Thunderbolt is obviously way faster (Macs don't support gen 2x2 unfortunately).

Ok, but does he have an M1 Mac like you do?

Don't get me wrong, the 5,1 was awesome. But I question the logic of putting more time and money into what's now a very old computer (that's stuck on Mojave without getting into OC). A stock 5,1 doesn't get 6000MB/s from an SSD, so unless he has an expensive Squid card, it's not really a relevant counterpoint.

I'm losing track of what the question is. I have a CmP 5,1 dual processor with 8TB SN850X as the boot drive. Consequently, it is fast for my edsiscovery use cases which include python3 / regex search work on multi-terabyte file sets. Per ChatGPT, I would need a pretty top spec Apple Silicon machine to equal the speed and/or utilize the drive, and have the graphics power to run the 49" 5120x1440 display, much less at HDR10 which the display is capable of but I'm not sure if I am getting now.

You clearly need the capacity to do your work, and it would indeed be extremely expensive if bought as an internal drive. A Thunderbolt enclosure for your 850x wouldn't be that much, though.

I doubt any Apple Silicon Mac would have trouble running that monitor - it has less pixels than a 4K display (they're just spread out more, over its large surface). Perhaps if you are looking to run it at 120Hz, but I doubt eDiscovery work requires that.

I do need a new laptop. And it would be nice to get one that could connect to the SN850X in an enclosure so I could continue to utilize it without a speed bottleneck. I believe it said that Thunderbolt 5 on the new Macbook Pro M5 Pro and M5 Max is theoretically that but that enclosures aren't available yet.

I'd look at some benchmarks for your software before getting too hung up on the maximum bandwidth of your SSD. It may be that other factors are far more significant, and whether you connect it via TB3, 4 or 5 has little impact on performance. This is certainly the case for e.g. graphics cards, where even a top end card will deliver 95% of its performance on a relatively limited slot.

But the bigger issue I need to solve is I want to keep using the new display, but I need to resolve the issue of Preview and Quicklook crashing my machine and requiring a power-off (button) restart all the time. The problem existed before when I was running a 32" 4k panel. But it was rare. With this new display at 5120x1440, it is not rare at all and it seems to be getting more frequent. Court documents are all in PDF. So PDFs are a critical part of my workflow.

I feel strongly that there *should& be a software fix for this. But I can't seem to find it, notwithstanding that there appear to be other reports of this happening to users with larger displays, and that the problem many not be limited to Mojave. https://discussions.apple.com/thread/255031336

I have no doubt that with enough effort, this problem is fixable. There must have been loads of Mac Pro users running (multiple) 4K screens. Maybe your 580 is on the fritz. Maybe upgrading to e.g. a 6600XT would help. Maybe installing OCLP and upgrading to Monterey would solve your issues. But especially for a machine that's used daily for legal work, is it really worth your time?

It seems like the main stumbling block to moving to a current system is Apple removing plugin support from Mail from Sonoma onwards. Personally, I'd investigate alternative email clients with better search features, or that play nice with HoudahSpot (e.g. MailMate). Aside from the various benefits of Apple Silicon, your OS would get security updates and you wouldn't be relying on 15 year old hardware to do your job.
 
  • Like
Reactions: AppleDrank
I'm losing track of what the question is. I have a CmP 5,1 dual processor with 8TB SN850X as the boot drive. Consequently, it is fast for my edsiscovery use cases which include python3 / regex search work on multi-terabyte file sets. Per ChatGPT, I would need a pretty top spec Apple Silicon machine to equal the speed and/or utilize the drive, and have the graphics power to run the 49" 5120x1440 display, much less at HDR10 which the display is capable of but I'm not sure if I am getting now.

I do need a new laptop. And it would be nice to get one that could connect to the SN850X in an enclosure so I could continue to utilize it without a speed bottleneck. I believe it said that Thunderbolt 5 on the new Macbook Pro M5 Pro and M5 Max is theoretically that but that enclosures aren't available yet.

But the bigger issue I need to solve is I want to keep using the new display, but I need to resolve the issue of Preview and Quicklook crashing my machine and requiring a power-off (button) restart all the time. The problem existed before when I was running a 32" 4k panel. But it was rare. With this new display at 5120x1440, it is not rare at all and it seems to be getting more frequent. Court documents are all in PDF. So PDFs are a critical part of my workflow.

I feel strongly that there *should& be a software fix for this. But I can't seem to find it, notwithstanding that there appear to be other reports of this happening to users with larger displays, and that the problem many not be limited to Mojave. https://discussions.apple.com/thread/255031336

But the direction ChatGPT has pushed in is that the crashes are due to Preview and Quicklook putting a special kind of strain on the video card, that while the XFX RX580 I've been using for the past 5-6 years in this machine is enough to power the display at 60 or 120 hz, it's not powerful enough to avoid window server crashing from Preview/Quicklook:
2026-05-25 20:16:34.883026-0700 localhost Preview[461]: (QuartzCore) Created Activity ID: 0x15500c, Description: Load Metal shader
2026-05-25 20:16:34.883052-0700 localhost Preview[461]: (QuartzCore) Created Activity ID: 0x15500d, Description: Load Metal shader
2026-05-25 20:16:34.883078-0700 localhost Preview[461]: (QuartzCore) Created Activity ID: 0x15500e, Description: Compile Metal pipeline
2026-05-25 20:16:34.884217-0700 localhost Preview[461]: (QuartzCore) Created Activity ID: 0x15500f, Description: Load Metal shader
2026-05-25 20:16:34.884228-0700 localhost Preview[461]: (QuartzCore) Created Activity ID: 0x155010, Description: Load Metal shader
2026-05-25 20:16:34.884232-0700 localhost Preview[461]: (QuartzCore) Created Activity ID: 0x155011, Description: Compile Metal pipeline
2026-05-25 20:16:34.884295-0700 localhost Preview[461]: (QuartzCore) Created Activity ID: 0x155012, Description: Load Metal shader
2026-05-25 20:16:34.884300-0700 localhost Preview[461]: (QuartzCore) Created Activity ID: 0x155013, Description: Load Metal shader
2026-05-25 20:16:34.884320-0700 localhost Preview[461]: (QuartzCore) Created Activity ID: 0x155014, Description: Compile Metal pipeline
2026-05-25 20:16:39.056608-0700 localhost kernel[0]: (IOAcceleratorFamily2) void IOAccelFenceMachine::fence_timeout(IOTimerEventSource *): AMDRadeonAccelerator prodding blockFenceInterrupt
2026-05-25 20:16:39.157599-0700 localhost kernel[0]: (IOAcceleratorFamily2) void IOAccelFenceMachine::fence_timeout(IOTimerEventSource *): AMDRadeonAccelerator prodding blockFenceInterrupt
2026-05-25 20:16:39.258599-0700 localhost kernel[0]: (IOAcceleratorFamily2) void IOAccelFenceMachine::fence_timeout(IOTimerEventSource *): AMDRadeonAccelerator prodding blockFenceInterrupt
2026-05-25 20:16:39.359646-0700 localhost kernel[0]: (IOAcceleratorFamily2) void IOAccelFenceMachine::fence_timeout(IOTimerEventSource *): AMDRadeonAccelerator prodding blockFenceInterrupt
2026-05-25 20:16:39.460681-0700 localhost kernel[0]: (IOAcceleratorFamily2) void IOAccelFenceMachine::fence_timeout(IOTimerEventSource *): AMDRadeonAccelerator prodding blockFenceInterrupt
2026-05-25 20:16:39.561666-0700 localhost kernel[0]: (IOAcceleratorFamily2) void IOAccelFenceMachine::fence_timeout(IOTimerEventSource *): AMDRadeonAccelerator prodding blockFenceInterrupt
2026-05-25 20:16:39.662666-0700 localhost kernel[0]: (IOAcceleratorFamily2) void IOAccelFenceMachine::fence_timeout(IOTimerEventSource *): AMDRadeonAccelerator prodding blockFenceInterrupt
2026-05-25 20:16:39.763675-0700 localhost kernel[0]: (IOAcceleratorFamily2) void IOAccelFenceMachine::fence_timeout(IOTimerEventSource *): AMDRadeonAccelerator prodding blockFenceInterrupt
2026-05-25 20:16:39.864239-0700 localhost kernel[0]: (IOAcceleratorFamily2) void IOAccelFenceMachine::fence_timeout(IOTimerEventSource *): AMDRadeonAccelerator prodding blockFenceInterrupt
2026-05-25 20:16:39.964318-0700 localhost kernel[0]: (IOAcceleratorFamily2) void IOAccelFenceMachine::fence_timeout(IOTimerEventSource *): AMDRadeonAccelerator prodding blockFenceInterrupt
2026-05-25 20:16:40.064379-0700 localhost kernel[0]: (IOAcceleratorFamily2) void IOAccelFenceMachine::fence_timeout(IOTimerEventSource *): AMDRadeonAccelerator prodding blockFenceInterrupt
2026-05-25 20:16:40.165384-0700 localhost kernel[0]: (IOAcceleratorFamily2) void IOAccelFenceMachine::fence_timeout(IOTimerEventSource *): AMDRadeonAccelerator prodding blockFenceInterrupt
2026-05-25 20:16:40.266377-0700 localhost kernel[0]: (IOAcceleratorFamily2) void IOAccelFenceMachine::fence_timeout(IOTimerEventSource *): AMDRadeonAccelerator prodding blockFenceInterrupt
2026-05-25 20:16:40.367371-0700 localhost kernel[0]: (IOAcceleratorFamily2) void IOAccelFenceMachine::fence_timeout(IOTimerEventSource *): AMDRadeonAccelerator prodding blockFenceInterrupt
2026-05-25 20:16:40.468407-0700 localhost kernel[0]: (IOAcceleratorFamily2) void IOAccelFenceMachine::fence_timeout(IOTimerEventSource *): AMDRadeonAccelerator prodding blockFenceInterrupt
2026-05-25 20:16:40.569439-0700 localhost kernel[0]: (IOAcceleratorFamily2) void IOAccelFenceMachine::fence_timeout(IOTimerEventSource *): AMDRadeonAccelerator prodding blockFenceInterrupt
(to infinity)
Ok so why not try my very 1st suggestions to squash your Quicklook issue?
 
I was referring to what @AppleDrank said. Basically - does he have a RAID card like yours, or is he just using a passive adapter? If the latter, he's only getting ~1500MB/s anyway, i.e. 4x PCIe 2.0. Even a cheap USB 3.2 gen 2 enclosure will give 1000MB/s, and Thunderbolt is obviously way faster (Macs don't support gen 2x2 unfortunately).



Don't get me wrong, the 5,1 was awesome. But I question the logic of putting more time and money into what's now a very old computer (that's stuck on Mojave without getting into OC). A stock 5,1 doesn't get 6000MB/s from an SSD, so unless he has an expensive Squid card, it's not really a relevant counterpoint.



You clearly need the capacity to do your work, and it would indeed be extremely expensive if bought as an internal drive. A Thunderbolt enclosure for your 850x wouldn't be that much, though.

I doubt any Apple Silicon Mac would have trouble running that monitor - it has less pixels than a 4K display (they're just spread out more, over its large surface). Perhaps if you are looking to run it at 120Hz, but I doubt eDiscovery work requires that.



I'd look at some benchmarks for your software before getting too hung up on the maximum bandwidth of your SSD. It may be that other factors are far more significant, and whether you connect it via TB3, 4 or 5 has little impact on performance. This is certainly the case for e.g. graphics cards, where even a top end card will deliver 95% of its performance on a relatively limited slot.



I have no doubt that with enough effort, this problem is fixable. There must have been loads of Mac Pro users running (multiple) 4K screens. Maybe your 580 is on the fritz. Maybe upgrading to e.g. a 6600XT would help. Maybe installing OCLP and upgrading to Monterey would solve your issues. But especially for a machine that's used daily for legal work, is it really worth your time?

It seems like the main stumbling block to moving to a current system is Apple removing plugin support from Mail from Sonoma onwards. Personally, I'd investigate alternative email clients with better search features, or that play nice with HoudahSpot (e.g. MailMate). Aside from the various benefits of Apple Silicon, your OS would get security updates and you wouldn't be relying on 15 year old hardware to do your job.
The logic is simple, he already invested into everything for a MP 5,1 so why not invest a little more to it on the cheap instead of a whole new machine? It doesn't have to be Squid brand, the quad pcie cards are cheap now and so are nvme blades.
PS - Catalina doesn't need OC.
 
Even a cheap USB 3.2 gen 2 enclosure will give 1000MB/s, and Thunderbolt is obviously way faster (Macs don't support gen 2x2 unfortunately).
macOS supports USB 3.2 gen 2x2 since Ventura. No Mac comes with USB 3.2 gen 2x2 controller though.
https://forums.macrumors.com/thread...-at-full-speed-on-a-mac.2443940/post-33591251
For Intel Macs, you can add a PCIe card.
For Apple Silicon Macs, you might be able to add a PCIe card in a Thunderbolt PCIe expansion chassis.

Alternately, you can get USB 3.2 gen 2x2 from a Thunderbolt 5 device that has a downstream Thunderbolt 5 port. This won't work with Apple Silicon Macs unless a Thunderbolt 3 device is placed between the Apple Silicon Mac and the Thunderbolt 5 device.
 
The logic is simple, he already invested into everything for a MP 5,1 so why not invest a little more to it on the cheap instead of a whole new machine? It doesn't have to be Squid brand, the quad pcie cards are cheap now and so are nvme blades.
PS - Catalina doesn't need OC.

There comes a point though where it doesn’t make sense to continue investing in an old computer. I’d argue we’re past that point for the 5,1, but it’s obviously up to individuals to decide. You’d surely agree that either way, it’s just putting off the inevitable? You can’t stay on an unsupported OS indefinitely, nor depend on aging hardware solutions.

macOS supports USB 3.2 gen 2x2 since Ventura. No Mac comes with USB 3.2 gen 2x2 controller though.

Noted, but this kind of makes it moot.


For Intel Macs, you can add a PCIe card.

Well, the Mac Pro specifically.

For Apple Silicon Macs, you might be able to add a PCIe card in a Thunderbolt PCIe expansion chassis.

Bit pointless though - might as well just use a Thunderbolt M.2 enclosure in the first place.

Alternately, you can get USB 3.2 gen 2x2 from a Thunderbolt 5 device that has a downstream Thunderbolt 5 port. This won't work with Apple Silicon Macs unless a Thunderbolt 3 device is placed between the Apple Silicon Mac and the Thunderbolt 5 device.

Worth knowing, but it seems Apple see USB as the cheap / fast enough option, with TB being the ‘FW800’ solution.
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.