Only bottlenecking if you have to use them - if there is enough VRAM to hold everything (like 24 GB!) then it's all good: x16 link PCIe 3.0 = 32GB/s
-
Competitor rules
Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.
NVIDIA ‘Ampere’ 8nm Graphics Cards
- Thread starter LoadsaMoney
- Start date
Only bottlenecking if you have to use them - if there is enough VRAM to hold everything (like 24 GB!) then it's all good: x16 link PCIe 3.0 = 32GB/s
It is a bit complicated - GPU-z can show the controller utilisation if the right sensors are exposed on your setup and with debug tools you can look at the data in use and get a rough measurement. I don't know of any way to directly measure it.
Battlefield 4 is a 7 year old game that lists a GTX660 as the recommeneded card on steam. I don't think it is a good/relevant example due to its age.
With the next gen consoles planning on utilising data streaming from the SSDs i do wonder when games are built to take advantage of that, will PCIE bandwidth begin to matter more.
It is a shame that nobody has done a proper investigation as to why HZD is behaving the way it is, it could give us insight into what to expect.
I find it funny people assume intel is holding back the Navi GPUs when Radeon themselves use a 9900k in there rest rig to post performance numbers every driver update. People have no idea 
RS-339, Testing by AMD Performance Labs as of August 13th, 2020 using a test system configured with an Intel Core i9-9900K CPU (3.6GHz), 16GB DDR4-3200MHz memory, a Radeon™ RX 5700 XT graphics card and Windows 10 x64 with Radeon Software Adrenalin 2020 Edition 20.8.2, and a similarly configured system with Software Adrenalin 2020 Edition 20.8.1, to test driver over driver FPS performance using the game A Total War Saga: Troy at 1920x1080 resolution. Performance may vary
RS-339, Testing by AMD Performance Labs as of August 13th, 2020 using a test system configured with an Intel Core i9-9900K CPU (3.6GHz), 16GB DDR4-3200MHz memory, a Radeon™ RX 5700 XT graphics card and Windows 10 x64 with Radeon Software Adrenalin 2020 Edition 20.8.2, and a similarly configured system with Software Adrenalin 2020 Edition 20.8.1, to test driver over driver FPS performance using the game A Total War Saga: Troy at 1920x1080 resolution. Performance may vary
I don't have anything much installed right now to test - but I had a quick look at it with some more recent games not that long ago and it was around 12% (usually lower*) on my setup - granted that is a 4820K @ 4.6GHz and a GTX1070 and not a 10K series Intel and a 2080ti
* In a lot of games it barely manages to register a percentage much of the time the odd spike aside - games tend to avoid shuttling data over the PCI-e link while rendering if they can avoid it - most of it is command traffic.
1 September is the fashion show
I say fashion because Nvidias founders cards always look better than AMDs
Associate
- Joined
- 12 May 2005
- Posts
- 1,782
All the tests prove is that the Intel 10900K system is slower at 4K for unknown reasons.
They should have tested PCIE3 vs PCIE4 on the same X570 3950x 2080ti system but they didn't.
They sound like there out dated even before they been released
They should have M.2 NVMe drives
Permabanned
- Joined
- 23 Apr 2014
- Posts
- 23,552
- Location
- Hertfordshire
They sound like there out dated even before they been released
They should have M.2 NVMe drives
PS5 has PCIe 4.0 SSD that can push compressed data at up to 9GB/sec, supposedly. Faster than NVMe and markedly faster than all PC drives, for now. (Compressed shenanigans though)
I've repeatedly spent thousands on hardware and peripherals too, I am quite prepared to sit this generation out and buy a console instead. I don't have to stop using my PC and I've got a decent enough PC for the games I'm playing now. My PC won't magically disappear if I buy a console and I can have the additional fun of being sociable at home with mates playing football or racing games. I'm certainly not going to overpay for hardware if I don't believe it offers value and I've owned Titan XP SLI, 295x2 Quadfire, 6990 Quadfire, 980TI SLI so it's not like I won't spend if I can see the performance I want from my money. If I can't see the performance e.g.2080ti+50% for £699 or thereabouts I'll wait and enjoy something else until Nvidia see sense. I was initially prepared to pay more than £699 but that's how much the 2080ti successor 'should' cost in my book.
Last edited:
5.5GB/s is uncompressed max speed. 9GB/s is average compressed and 20GB/s is the max compressed speed
As per Sony own admission though, the Theoretical max of 20GB/s with compression is very difficult to achieve hence wuothe 9GB/s average
Last edited:
Can I get a link to that Gamers Nexus article?
For now? You mean Ryzen 4000 right? I follow that the least of all pc part sectors i must take a look at this game nexus article too if someone will link.
But for sure i need a big overhaul the only good thing in my case is the AX860i and the 2080ti the SSD badly needs replaced soon its 256gb fs.
If you watch the video he does point out that intel (fps) are still better for gaming, the rest is just speculation.
Are you also aware that Horizon Zero Dawn has just had a couple of patches to fix performance problems.
Software that does not work correctly should never be used to benchmark anything.
I'm not sure the data read is correct unfortunately. If you look here you can see low pci-e utilization and yet the game has clear performance penalty at x8 vs x16. That would mean it's a very narrow bottleneck. There's just something screwy with this game but without a proper gpu profiler I doubt the data from these tools (gpuz etc) as well.