GPU 3DMark PCI Express Test & Riser cable

SashaLag

SFF Lingo Aficionado
Original poster
Jun 10, 2018
127
111
Hello guys,

given the introduction of the new PCI Express test by 3DMark, I was wondering if somebody of you may want to test the (probable) loss of performance introduced by a flexible riser cable (with and without)... It would be awesome (and purely accademic) to know how much a given model impact on performance of a PC. Alternatively, have somebody of you already found a website with this kind of comparison?


Thank you so much!
 

Scott

Caliper Novice
Nov 29, 2016
29
17
Here is a really good reference for PCIe 3.0 flexible risers


Basically, a quality shielded riser has little to no impact on performance. A cheap riser can have a very small or very big impact on performance, but it seems to be related to power delivery - cards with no external power like a GTX 1050 see big losses on cheap risers and very small losses with high-quality risers while cards with external power see much smaller losses with cheap risers and no measurable loss with high-quality risers.
 

Necere

Shrink Ray Wielder
NCASE
Feb 22, 2015
1,719
3,281
Speed testing is cool and all, but what I'd really like to see is a test that can detect errors in the data moving across the PCIe bus. In my (admittedly limited) experience testing them a few years ago, I found that a riser could seem to operate fine in one benchmark or game, but show subtle graphical artifacting in another if it was defective in some way. Those kind of difficult to detect errors were one of the things that put my off using flex risers in my case designs (at least those that could potentially make it to production). Also I believe there are some hard limits on trace length and number of interconnects that make risers nonviable for PCIe speeds in the future, with PCIe 4.0 and on.
 

SashaLag

SFF Lingo Aficionado
Original poster
Jun 10, 2018
127
111
Thank you to both of you! That comparison was very interesting to read!