News Intel: Rocket Lake's PCIe 4.0 Storage Performance is 11% Faster Than AMD Ryzen

Feb 24, 2021
43
8
35
0
Intel is trying to make 11% faster storage look like something huge like CPU performance difference... no one would care about 11% better read write, for Such differences matters in servers only , which this CPU is not aimed at to begin with.
 
Reactions: Soaptrail
Feb 24, 2021
43
8
35
0
For a long time, AMD's chipsets were ~10% worse than Intel's at USB and SATA. As evidenced by AMD's USB drop-out issues, AMD still has some IO quirks to work out and it wouldn't be too surprising if those affected its 4.0 performance.
NVME Storage on PCIe 4.0 has nothing to do with SATA or USB controllers . not related at all. This is direct CPU lanes , no bridge or chipset in between .
 
Reactions: Makaveli

InvalidError

Titan
Moderator
NVME Storage on PCIe 4.0 has nothing to do with SATA or USB controllers . not related at all. This is direct CPU lanes , no bridge or chipset in between .
So what? PCie lanes don't directly connect to CPU cores, they have to go through a routing and arbitration layer that interfaces with the infinity fabric between the IOD, CCD(s) and between cores just like they do between the uplink and downlinks within a chipset. Glitches and excess latency at any stage can cause substantial performance degradation and AMD's Zen 2/3 start with a ~20ns handicap courtesy of its CCD-IOD interface.
 
Feb 24, 2021
43
8
35
0
So what? PCie lanes don't directly connect to CPU cores, they have to go through a routing and arbitration layer that interfaces with the infinity fabric between the IOD, CCD(s) and between cores just like they do between the uplink and downlinks within a chipset. Glitches and excess latency at any stage can cause substantial performance degradation and AMD's Zen 2/3 start with a ~20ns handicap courtesy of its CCD-IOD interface.
Had it been the case we would have seen 11% GPU decrease in performance as well.
 

Makaveli

Distinguished
Jan 15, 2001
1,013
142
19,470
4
I am taking these Intel internal benchmarks with a pinch of salt. Independent testing will either confirm or deny this claim when these CPU's, motherboards and RAM appear on the market.
Agreed.

Wait until a full review is out.

And that fact that this data is by Ryan shroud who is a major shill I would wait for more information.
 
Last edited:

InvalidError

Titan
Moderator
Had it been the case we would have seen 11% GPU decrease in performance as well.
Increased latency and reduced bandwidth don't have anywhere near 1:1 correlation with graphics performance: look at PCIe scaling benchmarks where going from 4.0x8 or 3.0x16 to 4.0x16 only yields a 0-5% difference for 100% more PCIe bandwidth on GPUs with enough VRAM to keep assets on-board.

As joker wrote, latency and bandwidth only affect how fast you can read/write from/to the GPU, not how fast the GPU does whatever it needs to do once tasks get in its job queue. PCIe latency and bandwidth has negligible effect on GPU performance once assets are loaded, which is why mining rigs are managing perfectly fine with PCIe 2.0x1 per card no matter how powerful the card is.
 

Soaptrail

Honorable
Jan 12, 2015
190
45
10,620
1
Who cares? Few will ever notice even IF Intel's claims stand up to scrutiny.
Agreed, no one would notice this in real world scenarios except maybe heavy video editing but even then someone would only know with benchmarks. But it is amazing you have to get a riser card to get the full bandwidth from Intel.
 

everettfsargent

Reputable
Oct 13, 2017
83
15
4,535
0
So no so-called apples-to-apples comparison, e. g. AMD Ryzen™ 7 5800X vs i9-11900K. I am thinking it is due to the differences in chiplets for AMD 5000 CPU's (8-core vs 16-core or 1 chiplet vs 2 chiplets). Don't have a clue about this benchmarks behavior(s) across different CPU layouts/cores. It would really be funky town if a 3800XT bet the i9-11900K in a PCIE 4.0 storage bench though. Oh and the same tests run through M.2 storage devices.

It could also be simply due to the latency caused by a larger physical distance between the CPU chiplet(s) and the I/O die.

Also, why use mediocre PCIE 3.0 boot drives? And who cares about a 3090 GPU in a storage bench as nothing should be GPU bounded (except if more test results are forthcoming).
,
 
Last edited:
Feb 24, 2021
1
3
15
0
Random thoughts:
  1. In my experience AMD may be a little behind in synthetic benchmarks but that is the price you pay to be first to market. If memory serves, AMD was first with SATA 3 on Bulldozer and those benches lagged by about 20-30% in my tests between AMD and Intel - I can not remember what OEM or boards, just that the configs were similar and both were fresh installs and newest drivers at the time of test. USB 3 throughput was a bit less as well.
  2. Intel damn well BETTER be 10 % faster since they have 10-50x the R&D budget and resources and have had the benefit of over 18 months additional time on their hands to get this done.
  3. Still, I wonder if this advantage only exists with one product from one OEM or is it large enough to measure across the board. What if I am reusing my 860 EVO? (as I did) Is it even noticeable then? Does this same performance exist in lower spec boards and CPU's or do I HAVE TO BUY a $500 board AND the top of the line CPU + 32GB RAM to get it? What if I get a mid tier ASROCK board along with a simple Quad I-5 and 16GB - as is my custom?
  4. How tweaked were their settings? It's not impossible to eke out an additional 10% using every trick in the book under the hood.
  5. I only ask because we know from previous experience that intel is not above a little sleight of hand in theses demos - didn't they pull a CES stunt demo with a commercial freezer under the table a couple years back to demo their ability to hit 5ghz? memory escapes me at the moment. I may not have the details right - just that there was a commercial grade freezer that you could not run off a normal 1500 watt residential 110 volt power outlet driving the show under the tablecloth.
But yeah - 10% in a synthetic bench on Gen IV SSD's? YAWNNNNNNNNNNNNNNNNNNNNNNNNN. How's it feel to finally get a desktop CPU on 10nm after 7 years of trying?
 
Last edited:

watzupken

Prominent
Mar 16, 2020
287
100
770
1
I don't know if Intel is still limiting PCI-E 4.0 to their Z series board and i5 and above CPU. If so, then the higher speed will not mean anything to most who may not want to invest in a Z series board. Current boards are also going in the way of the dodo with Alder Lake coming on a new socket.
 

Giroro

Honorable
Jan 22, 2015
776
170
11,190
13
If Intel had a more meaningful benchmark win to show off, then they would have shown them off.
The only people who care about this kind of minor storage bandwidth difference are the ones using heavy sustained workloads over a long period of time... If Intel had a significant lead in those benchmarks, then they would have shown that instead.

If photoshop launches in 5 seconds instead of 5.5 seconds its like... Alright. Neat.
But ultimately windows, the filesystem, and software are the bottlenecks right now so the real life difference may not even be measurable, let alone noticeable. I barely even noticed a difference when moving up from an old ~300MBps sata SSD to a nvme drive promising roughly 10x the rated performance.
For what it's worth, that system with an i5 4670k and sata SSD booted significantly faster than my shiny new ryzen 3700x did on nvme. Because, the storage bandwidth wasn't the problem. It was Ryzen's slow buggy launch BIOS, which still have some room for improvement.
 
Reactions: Redneck5439
Feb 24, 2021
43
8
35
0
For what it's worth, that system with an i5 4670k and sata SSD booted significantly faster than my shiny new ryzen 3700x did on nvme. Because, the storage bandwidth wasn't the problem. It was Ryzen's slow buggy launch BIOS, which still have some room for improvement.
This is wrong comparison , you need 1 to 1 hardware on the motherboards to compare booting speeds between two PCs. this includes everything from LAN chips to sound chips and even chipset integrated functions .
 

Redneck5439

Honorable
Aug 21, 2015
1,275
66
11,740
175
For a long time, AMD's chipsets were ~10% worse than Intel's at USB and SATA. As evidenced by AMD's USB drop-out issues, AMD still has some IO quirks to work out and it wouldn't be too surprising if those affected its 4.0 performance.
No doubt, as amazing as the Zen 3 processors are, AMD still has some improvements to make/ bugs to work out. I absolutely love my 5900X, and have managed to get very nice overclocks out of it (very nice for Zen arch that is). From the benchmarks I have seen the new 11900K will have nearly the same single core performance and get stomped in multi-core performance (I score the same ~700 single core in CPU-Z, get better single core in CB R20 of ~650, and loose in Geekbench where my best single core score ~1800). In multi-core situations the 11900K is totally out of its league as the 5900X has 4 more cores and 8 more threads, so there's no point even going there.

Something I have noticed when benchmarking my system is my external WD Elements hard drive will disconnect under heavy load and then reconnect to the system. This typically causes my antivirus (Kaspersky total security) to quick scan the drive when it reconnects and is more than likely effecting some of my benchmark scores. Its not a huge problem, and thus far has only happened when the processor is under a full sustained load, but it is there and it is none the less annoying. I also have to worry that it might happen when doing a data transfer from my internal drives to my external storage as I will probably loose data / projects if it were to occur. Thus far I have moved probably a terabyte of data back and forth since upgrading to the Dark Hero and 5900X without issue, but it remains a concern.
 

ASK THE COMMUNITY

TRENDING THREADS