The Sneaky Thing About PCI Express - CPU vs. Chipset

How do I determine this for my legacy Apple computer CPUs? Where can I find out if what if it supports PCIe 1.0, 2.0, 3.0, or 4.0? I have Apple Powermac G3, G4, and G5s as well as Mac Pro 1,1 2,1 3,1 4,1 and 5,1.

Of course if you have Intel the graphics are on the CPU and run at the same speed . No PCI latency or bus problems.

Thanks for the video!

Man! I still use PCIe 2.0 x 1 😢

soooooo if i have pcie thst has a bandwidth that is enought for my gpu in pcie by 8 is that enought than and can i put in two gpus?

Why have I only just found out about this after putting together my own rigs for 9 years, branding around this is terrible especially from the motherboard manufacturers

N THIS EPISODE. MISTER HARRY POTTER DEW FUNNY FACE IN THIS EPISOND LOOKED AT THE FACE AT THE HOST IN THE STORY AND EYES TO TO GINNY WEASLEY HE. LOVE

N THIS EPISODE. MISTER HARRY POTTER DEW FUNNY FACE IN THIS EPISOND LOOKED AT THE FACE AT THE HOST IN THE STORY AND EYES TO TO GINNY WEASLEY HE. LOVE

Senario, Questions Bellow.
Z790 Chipset
13900K
RTX4080
Max# of PCIe Lanes = 20
16 lanes are used for GPU
4 lanes left to the CPU.

Question
How will the system suffer if i use 2 NVMe storage devices running in RAID 0 if both connections are to the CPU instead of the chipset. Will the SSDs run at 2 x2 speed, esencially spliting the 4 lanes into 2 lanes for each drive?

Question
Should I instead use one NVMe SSD as a boot drive and have a RAID 0 array connected to the chipset for storage?

Quesiton
Correct me if I'm wrong but the Z790 chipest has a x8 lane to the cpu, Should I just use RAID 0 on the chipset and leave the extra 4 lanes open for the CPU?

So dumb question, but let's say I have a system with 16 PCIe lanes. All of which are being used by a Nvidia 1080 and a have one NVMe which is installed directly on the motherboard. If I use the second nvme slot, or even install a third through a NVMe adapter, is there any positive benefit going that route, as opposed to using multiple SATA SSD's? Or will it just introduce more problems and knock my GPU lanes down? I know many people don't see the speed benefit in NVMe but I have in a few instances when transferring large amounts of data.

Is that a pot plant in the background?

i belive that's the reason why some risers cards dont work on certain m.2 pci lanes.

As standard user, medium game, lot in editing video, 2 nvme storage. do I need dmi 4.0x8 instead of dmi 4.0x4? This is b760 x z790. Thank you

XD when ppl buy into AMD for its "PRICE ORIENTED" implications....then populate 1 too many sata ports and PCI-e lane on cpu gets cut to 8x on gpu bc cpu is stipened by a large pci-e population VIA SATA OH LUL...lemme gues quick solution buy a more expensive amd board that "WORKS" LUL

So, when i have 2 SSD drives and one gpu on The top slot, should i ser pcie source pch or cpu? I’m lioking for The best performance

I knew this

If my CPU ad 20 lanes, then my GPU get's direct access but my M.2 Drive still routes through the chipset right? Coz the oter 4 lanes coming off the CPU is for downlink to the chipset? So If I had 16 lanes CPU then my x16 slot wouldn't even be able to be fully utilised.

Someone correct me pls.

Does this mean it is not possible to run a gpu at x16 with the use of an m.2 connected via cpu pcie lane with an intel 10850k?

CPU vs PCH

About the mouse you mentioned at the end. I use a Bluetooth mouse with my laptop for the convenience of not needing a dongle but I don't really see much if any latency. It is one of those things most people might not even notice really. I have an HP XB4000 mouse and an Intel AX200NGW that I use for Bluetooth and WiFi. That card can be gotten cheaply but I mainly use it because it is a good one. It is not a tri-band card but it is WiFi 6 and I get a good fast connection 40 or 50 feet away on the other side of the house. This place is not small but a good TP-Link router seems to cover it well and without a horribly slow connection too! This was more about a mouse than anything and quickly expanded into other things.

Free info thanks linus

Is connecting devices directly to the cpu taking Performance of the cpu instead of the chipset? So can it be good to connect devices to the chipset?

generation

Good Information! Can you tell how the integrated graphics of CPU (SoC) is connected to the CPU? Do they also consume some lanes of PCIe? For example Intel CPU with integrated graphics.

So if I heard you correctly, the M.2 takes up 4 lanes of the PCIe band width? Clearly Intel needs to have more PCIe lanes to their chip.. If I recall some older chips had 40 if I am not mistaken, like I have with my 4930k.. so why are we down to 20 now days? Seems like we need more PCIe lanes to me available through the CPU.. and the very fact that the M.2 drives are being ran through the PCIe lanes, should necessitate that INTEL start to increase these lanes on their CPU's...

Thanks for the nice breakdown. What exactly is a "lane"? Is it one pin coming off of the CPU? I also don't understand how these speeds are measured. What does a speed of 8x or 16x mean? what is the "x" unit measuring? If it's a multiplier as in "x times", what is it multiplying? In other words, it's 16 times what?

All the PCI 3/4 X??? lark is a stupid idea and they should make the manufacturers only produce the highest speed version. I mean how stupid to create a specification and then hobble it with crap components.

SATA, AHCI, NVMe, PCIe, M. 2 all these just confuse me a lot😐

So my i9 9900k can only read 24 PCIe lanes, will that said when i crossfire or use SLI nvidia, with 2 pcie x16 will i loss 8 lanes and even more when i use other pcie ports as x4 to my sound card

Almost makes you miss a North Bridge.

" H670 will have up to 12 PCIe 4.0 lanes, and up to 12 PCIe 3.0 lanes. B660 will have up to 6 PCIe 4.0 lanes, and up to 8 PCIe 3.0 lanes. " Can someone tell me What does it mean ?

Does the nvme mess with how many lanes my gpu run on?

Very informative, bless you Linus for continuously keeping us up to date in a language average users can understand.

Thnk u for the nice video. On my older computer with an asus motherboard Ramage extreme ii (pci-e 2.0 x 16 slut) I just installed a pci-e to usb 3 card (the LC-PCI-usb 3.2). I use a samsung t-7 drive. Now it starts transferring great even at around 750 mb/sec but after around 6 secs it drops to 35 mb (which of course doesnt make much sence for me because I work with big video files). How ever, from the comp back to the drive it does transfer around 105 mb/sec. What am I doing wrong? How can I ghet at least the 100 mb/sec?

When did your face decided that it could grow a beard? Did COVID give you a beard or is that just your wife clinging to your million dollar nonsense videos about unicorn tech face???

I'm still confused. If AMD maxes out at 24 pcie lanes, four are taken by the CPU and 16 by the motherboard, does that mean that using a NVME 4.0 m.2 ssd uses another four and maxes them out?

How do people put capture cards, network cards and other peripherals in their PCIe slots? Do they just stick to the older 3.0 standard for their m.2?

Can anyone help me with this? I have an ASRock B450 pro4-f motherboard with r5 2600x and a msi rx 570 8gb gpu. I have no SSD just two HDD. But for some reason my gpu is running at x8 3.0 instead of x16 3.0. I'm just hoping nothing is faulty here. And yes I have put the gpu in the 16x slot.

finally i appreciated you after this video

Just buy an IBM mainframe; The smallest configuration has 160 PCIe 3 lanes. They all connect directly to the CPU chips :-)

What I really want to know is Should we REALLY Turn off PCI express ASPM?

will it really make a REAL difference when it comes to gaming performance?

Every instruction about power management goes for "TURN this OFF ITS BAD next TURN THIS OFF IT IS ALSO BAD " but how bad is it actually?

It seems like someone spread a rumor about PCI E.ASPM, and every other person keeps saying the same thing assuming this is the RIGHT, but can ASPM make a BIG difference in gameplays after it exit one of the states? which are:


L0S (concerns setting low power mode for one direction of the serial link only, usually downstream of the PHY controller).
the moderate state

while

L1 (shuts off PCIe link completely, including the reference clock signal, until a dedicated signal (CLKREQ#) is asserted, and results in greater power reductions though with the penalty of greater exit latency.)
for maximum power reduction


it clearly can reduce power consumption, But is it possible that they will returned to L1/L0S during gameplays( or any activity)?
This for sure would be reasonable to turn it off, but I don't think they will be entering and leaving L1/L0S states during gameplays, so WTF do we need to turn it off. Does it reduce FPS?



If anyone can give me a good explanation I would love to hear you. The only thing I can find in the internet is what everybody already knows TURN IT OFF and be happy. But I really want to know the reason why;

Bro , I have a latest gaming motherboard and there is three pciex1 slots, two are standard sizes small as pcie×1_1 and pcie×1_2 and the last one in the longest named as pcie×1_3. Does pcie×1 sizes may vary?

Can anyone identify the type of tree in the background? I need that for my bedroom. :-)

I still don’t understand what’s better CPU or CHIPSET someone help plz

This is a great video, very informative, I have a query I hope someone can help with though as this video doesn't quite answer it. My CPU has 16 lanes, My chipset has 24, I have a GPU in x16 AGP slot so this uses all the CPU lanes, If I connect a M.2 drive in a x4 slot managed by the chipset, How does the Chipset communicate with the processor if the processor is already using all avaialble lanes communicating with the graphics card? Does the GPU drop down to x8? This is what I fear, I'd like to have a m.2 but don't want to halve my graphics bandwidth. Any Help appreciated.

Bit of a wildcard question but trying to understand as a newbie - I've got a razerblade 15 2018 and just got 2 m.2 nvme pcie 1tb drives with usb 3.1 enclosures with the idea to record audio to one, whilst simultaneously reading samples audio files from the other (inside a daw) and using the main computer nvme drive to run the program. Im racking my brain trying to figure out if something like a Nestor NA611TB3 (dual slot thunderbolt 3) would be faster than 2 individual enclosure each plugging into a 3.0 slot in the computer.

Apparently the razerblade TB3 port has full 4 PCIe 3.0 lanes, but it's through the PCH and the processor is the Intel(R) Core(TM) i7-8750H CPU @ 2.20GHz, 2201 Mhz, 6 Core(s), 12 Logical Processor(s). Would there be performance issues to read and write at the same time through the same TB3 slot? Or would it be better to plug each enclosure in a USB A 3.0 input in the computer. Im assuming these USB A ports (3 in total) go through one lane regardless so its capped at that speed? Also one Thunderbolt enclosure and 1 cable would be more compact. Any help would be really appreciated!

the real slim shady ..

PCIe 4 now standard on most new motherboards.

So if I got the god awful 11900k, I should be ok with One 4th gen m2, two 3rd gen m2's and one 3090 without any potential slowdowns or bottlenecks? If I'm understanding this right