Question about PCIE Gen Setting
Printed From: ASRock.com
Category: Technical Support
Forum Name: Intel Motherboards
Forum Description: Question about ASRock Intel Motherboards
URL: https://forum.asrock.com/forum_posts.asp?TID=3283
Printed Date: 27 Sep 2024 at 6:57pm
Software Version: Web Wiz Forums 12.04 - http://www.webwizforums.com
Topic: Question about PCIE Gen Setting
Posted By: Souths1der
Subject: Question about PCIE Gen Setting
Date Posted: 19 Aug 2016 at 11:53pm
|
I'm having an issue with a riser cable. While I know it's an issue with the riser cable, the solution to make it work is changing a setting in the BIOS. I'm curious what the ramifications are in the setting I need to make my video card work. I have Thermaltake P5 case and I bought it because I like the looks of the video card being mounted vertically. To do that you have to use a riser cable. The following issue happens with their supplied riser (a known issue at Thermaltake) and a high quality 3M riser I purchased while troubleshooting. With the riser plugged into PCIE1 and the BIOS setting for that slow set to Auto, Gen3, or Gen2 I get no video from my card. My card only works with the setting at Gen1. This problem exists with both the above cables and two different video cards (R9 390X and GTX 1070). I know that the settings probably have no effect on gaming performance, etc. I've read all those tests. However, I also do some overclocking, and need to get every last clock. So I'm curious what is actually happening to the communication through that slot with the setting at Gen1. Is it no longer communicating at x16 but at x8 or x4? Is it still running at max Gbps? This information is what I can't seem to find anywhere. Thanks in advance for any info you can provide.
|
Replies:
Posted By: wardog
Date Posted: 20 Aug 2016 at 5:12am
|
An even better solution, and free, can be found at .... http://community.thermaltake.com/index.php?/topic/43213-attn-users-with-thermaltake-riser-cables/" rel="nofollow - http://community.thermaltake.com/index.php?/topic/43213-attn-users-with-thermaltake-riser-cables/ |
Posted By: Souths1der
Date Posted: 20 Aug 2016 at 5:56am
|
I don't mean to sound rude, but this doesn't answer my question. I'm well aware of the replacement program, some users still have issues with the replacement cable. That's why I went with an arguably better after market cable which also did not solve the issue.
|
Posted By: Xaltar
Date Posted: 20 Aug 2016 at 6:07am
|
PCIe Generations in essence doubled bandwidth from one to the next so: PCIe Gen 1 x16 = PCIe Gen 2 x8 = PCIe Gen 3 x4 So by running in Gen 1 mode you are basically running at PCIe Gen 3 x4 bandwidth. There should be no problem with the R9 390X but I can't say for certain if it will bottleneck the GTX 1070 or not. I would hazard a guess and say it shouldn't. I do not have a 1070 or 1080 to test with so I can't be 100% certain. I do know that prior to the release of the GTX 10XX series there were no GPUs capable of fully saturating PCIe gen 3 x4 bandwidth. I hope this is what you were after.
------------- 
|
Posted By: wardog
Date Posted: 20 Aug 2016 at 6:13am
|
I don't mean to sound rude but how can a BIOS setting make up for the myriad reports of this riser issue when the real problems lie with the graphic cards PCB and power? Hence why Tt is asking for such detailed system specs for replacement. If it was as simple as just mailing out ONE replacement cable they wouldn't need all that info. Just the reciept of the P5 provided. NM Tt aren't having much luck with correcting this for AMD/Radeon cards. Simple? Not really. |
Posted By: Souths1der
Date Posted: 20 Aug 2016 at 8:55am
You can try to come off as superior all you want (it is the Internet after all), but what is "simple" is that you just didn't answer my question. That's what I was trying to get across to you.
Regardless of your opinion on the value of the answer to my question, I did ask a specific question. I did not ask you to "solve" my problem. If you don't know the answer or can't add value toward that effort, then just stay out of it. |
wardog wrote:Posted By: wardog
Date Posted: 20 Aug 2016 at 9:48am
|
IT IS NOT A BIOS ISSUE |
Posted By: Souths1der
Date Posted: 20 Aug 2016 at 10:41am
|
What do you not get? Using all caps doesn't make you look smarter, it makes you look silly because you're not even paying attention to what I wrote in my first post. I agree with you, I'm not having a BIOS problem, never even hinted that I did. It's a cable issue, in that the cable somehow is not able to let me run the slot in Gen3 mode (no matter what cable I use). And again, let me repeat myself, I'm not asking for a solution. I asked a specific question about what happens to the slot when you put it in Gen1 mode. Thankfully, the question was answered by another user who does know how to read for comprehension. We can all go back to our regularly scheduled program now.
|
Posted By: wardog
Date Posted: 20 Aug 2016 at 10:55am
|
You could see if a powered x16 riser might do whatever needs done. http://www.newegg.com/Product/Product.aspx?Item=9SIAAD046B4566&cm_re=Pci-e_riser_cable-_-9SIAAD046B4566-_-Product" rel="nofollow - http://www.newegg.com/Product/Product.aspx?Item=9SIAAD046B4566&cm_re=Pci-e_riser_cable-_-9SIAAD046B4566-_-Product Sometimes that Tt thread seems to be power related, other times you could take away it as a signal related issue. re:signal We're talking small voltages and, well, a ribbon cable isn't exactly the most shielded. NM the added length the signal travels. Yet nobody is reporting any issue when their cards are plugged directly into a motherboards PCIe x16 slot. |
Posted By: Xaltar
Date Posted: 20 Aug 2016 at 1:09pm
|
Wardog is simply trying to provide some insight into the problem that brought you here. I provided an answer to your question. Risers have been around a long time, many of which are old and only able to work on a certain PCIe Generation. Some require gen 2 others gen 1. In some cases they simply have compatibility issues with certain motherboards, in many of these cases using a powered riser as Wardog mentioned proves to be the solution. To answer your question again: If you are using an x16 riser on the x16 slot and have the slot set to Gen 1 then you are operating at the equivalent of PCIe Gen 3 x4. If you are connected to an x8 slot then it would be equivalent to PCIe 3.0 x2 and PCIe Gen 3 x1 if it is connected to an x4 slot. This is because: PCIe Gen 1 = 250MB/s bandwidth per lane PCIe Gen 2 = 2x PCIe Gen 1 bandwidth PCIe Gen 3 = 2x PCIe Gen 2 bandwidth = 4x PCIe Gen 1 bandwidth See http://www.guru3d.com/articles-pages/pci-express-scaling-game-performance-analysis-review,2.html" rel="nofollow - this breakdown for more info. If you look at the benchmarks you will note there is very little difference between the 3 generations in actual gaming metrics. We are talking at most 3 - 5fps between gen 1 and gen 3 on a single GPU system. This could be considered within the margin of error for these benchmarks. The more telling evidence is the frame variance graphs which show all 3 tests pinned almost exactly on top of each other. If there was a bandwidth bottleneck then there would be sharp spikes peppered along the graph where this was happening. The reason for this is that current GPUs (at the time of that review) were not capable of fully saturating PCIe Gen 1 x16's 4000MB/s bandwidth. Now if you look at the little table above you will note that that equates to the same throughput as PCIe 3.0 x4 or PCIe 2.0 x8. Now without bandwidth details on the GTX 1070 I can't say if it will be more or less effected. On one hand it is a significantly faster GPU than the GTX 980 used in the review I linked but on the other it almost certainly uses more advanced texture compression and other algorithms. This means it could go either way, it could use more or less bandwidth than the previous generation depending on how Nvidia optimized it. My guess is that it will be more or less the same.
-------------
|
Posted By: parsec
Date Posted: 20 Aug 2016 at 8:07pm
IF you are still wondering about this, which has been answered above, at least IMO it has, I'll repeat it in my words. The answer is bandwidth or speed of the interface is lower in Gen1 mode (PCIe 1.0), than it is in Gen2 mode (PCIe 2.0) or Gen3 (PCIe 3.0.) So no, it is not still running at the maximum Gbps in Gen1 mode. Given this, the affect on gaming performance is obvious, up to a point. Normally a Gen3/PCIe 3.0 x8 connection will be more than any video card can use. With new video cards, 4K displays, and VR, that may no longer be true. The number of PCIe lane connections, x16, x8, etc, is not changed at a Gen1 setting, or any setting of that option. The need to set the PCIe connection to Gen1/PCIe 1.0 speed when using the TTake riser cable seems to indicate it is so cheaply made, or were defective from the factory. But since the 3M cable did the same thing, it makes me wonder. The thought of the CPU trying to maintain Gen3 bandwidth through a riser cable, which likely more than doubles the length of the connection to a video card, does not inspire confidence IMO. I forget what board you are using, and I don't see that information in this thread. ------------- http://valid.x86.fr/48rujh" rel="nofollow"> 
|
Posted By: Souths1der
Date Posted: 21 Aug 2016 at 1:24am
|
Z170 OC Formula. Through my research the added length comes up every now and then. I wonder if that is what the problem is. The real problem is my apparent OCD. I know this won't affect gaming to the human eye, and it should not affect benchmarks but by a few marks. It's the knowledge that I have this great board, and now a great video card, but their connection is not set at the best possible.....and it's not set simply because of appearance. Thanks to everyone.
|