chaosmaou

My new riser cable arrived yesterday and I installed it. Windows functions normally, and I ran a bunch of 3DMark tests to test for stability/performance of the new replacement cable sent to me directly from Thermaltake under warranty. I got the following scores in 3dMark scores during my testing. (You can click the links below to see the verified scores on 3dmarks website if you wish). The highest scores are in bold. These results result in a 3 to 2 win for the new riser cable (within the margin of error, performance is identical.) 3DMark Fire Strike (1080p) Without Riser: 16,843 3DMark Fire Strike (1080p) With Riser: 16,924 3DMark Fire Strike Extreme (1440p) Without Riser: 9,715 3DMark Fire Strike Extreme (1440p) With Riser: 9,701 3DMark Fire Strike Ultra (2160p) Without Riser: 5,295 3DMark Fire Strike Ultra (2160p) With Riser: 5,299 3DMark Time Spy (1440p) Without Riser: 6,887 3DMark Time Spy (1440p) With Riser: 6,864 3DMark Time Spy Extreme (2160p) Without Riser: 3,180 3DMark Time Spy Extreme (2160p) With Riser: 3,175 When I ran the tests without the riser cable installed, I had nearly 24 hours of uptime before starting the tests, and I had 0 pci errors reported by nvidia-smi after these tests were ran. HOWEVER - with the new cable installed, I had 507 pci errors, and my system had only had 37 minutes of total uptime once the same tests were completed! Errors reported by nvidia-smi are single bit CORRECTABLE errors, meaning that the data needed to be re-transmitted. Any double bit errors are the type that cause game crashes and instability that can lead to blue screens and random reboots. These kinds of errors may or not show up in the counter, depending on their severity (if the error causes a blue screen or reboot for instance). These cables clearly have what appear to be signal-to-noise issues (the foil trick on the old cable reduced the errors by 10x, see my previous post), but the new cable isn't causing any direct harm to system or game performance at this time. After a full day I haven't noticed any frame drops, and have had zero game or system crashes. In 2023, I would not recommend buying the cable new (AC-053-CN1OTN-C1) and using it. Thermaltake sells this pci-e 3.0 riser on their amazon store for $34.99. Their "premium" pci-e 3.0 cable costs $86.11, and their PCI-E 4.0 one is $89.99 at the time of this post, making their premium 3.0 cable incredibly overpriced. Both the newer cables are 10cm longer than the original one as well.

I recently starting having this issue after 1 1/2 years of use with the cable that came bundled with my Core P3 case. I still run an older system with a 7700k and GTX 1080 8GB. The issue happened right after rearranging my room, so I didn't suspect the cable for a long time, but removing it fixed my crashes and freezes. With the riser cable used I get 50-60 PCI errors reported from nvidia-smi every hour. Mounting my gpu directly to the board I get 0 in 3 days. You can use the following command from an admin powershell window assuming you have fairly recent nvidia drivers installed: nvidia-smi dmon -s pucvmet -d 1 -o DT This will show your GPU power usage, utilization, temps, and other info. The last three columns are the most important -- they show pci errors as well as your receive and transmit across the PCI link. The "pci errs" column is a counter -- it only resets after a full system reboot, or when a reset command is sent via nvidia-smi. I noticed some people in the thread wrapping their riser cables in aluminum foil to try to block RF interference. I was curious and gave this a try, using 2 whole layers around the cable and then wrapping it in clear non-conductive packing tape for safety. Surprisingly it reduced my PCI errors by over 10 fold. I only got 38 in a 24 hour period with the cable compared to more than that in a 1 hour time period previously. Removing the foil while the system was running save me another 20 more PCI errors within 30 minutes. This is for sure an RF interference issue with these cables and appears to not be the result of failure. It is possible that something in the cable that might help with RF has failed, but I don't think there is anything like that in PCI-E 3.0 cables (no repeaters like the newer 4.0 cables). Rearranging my room and putting my PC into a higher positioning away from the floor must be causing it to receive more signals, possibly from my AX wifi router or other devices. WARNING: I do NOT recommend wrapping this cable in foil -- either aluminum or copper. This can cause inductive capacitance, and is much more possible on more modern and powerful hardware because of the jump in amperage being used. There is a high chance of destroying your hardware when you do this, especially if the cable rubs against anything to ground, as the foil could cause shorts. I did this to experiment on a much older system. Aluminium foil is not as good as a RF blocker that people think. One can buy real copper RF tape online, but I previously talked to thermaltake support and they are mailing me a replacement riser cable. I will retest with that replacement cable when it arrives (which I assume will be the exact same cable since it's replaced under warranty) to determine if it a failure issue, or an overall build issue with these style of cables. If the new cable has the same RF issues, or whatever result I get, I will report my findings to help others. When I build my new system I will probably end up buying Thermaltakes premium PCI-E 4.0 cable straight 30cm, as it's the only cable that I know will mount to my Core P3 case without modification and is the only cable I can find in stock online (their 20cm cable isn't readily available on amazon for instance).