Gen5 NVMe have active cooling

Page 6 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

thecoolnessrune

Diamond Member
Jun 8, 2005
9,673
583
126
From my PoV I just want something serviced with enough PCIe lanes that I don't have to worry about whether the PCIe slot my video card is in going down from 16x to 8x because I put in too many drives. That or if there aren't enough lanes in total for all the drives I've installed, I'd want them to be shared instead of stealing speed from the video card slot.
Yeah with NVMe when it comes to sharing the only option is either a PCIe switch, which enthusiast boards have seemingly all but stopped doing, or an NVMe eHBA like the Broadcom 9600-24i that can natively attach 6 NVMe 4 Lane drives back to a single x8 PCIe port. More than a couple NVMe drives is still in my opinion beyond an enthusiast class need because it sucks up available PCIe lanes so quickly.

It would be cool to see an enthusiast enclosure for say up to 8 drives back to a single PCIe attachment. The technology certainly exists. Microchip's PFX 28xG5 PCIe Fan-out is basically built for that sort of thing, but at over $300 per chip in small quantities I think it will be a long time before we ever see these built into something like Icy Docs. They're mostly relegated to server backplanes for All-NVMe Storage when the host doesn't have enough lanes in the design for direct-to-NVMe.
 

DrMrLordX

Lifer
Apr 27, 2000
22,184
11,890
136
Yeah with NVMe when it comes to sharing the only option is either a PCIe switch, which enthusiast boards have seemingly all but stopped doing, or an NVMe eHBA like the Broadcom 9600-24i that can natively attach 6 NVMe 4 Lane drives back to a single x8 PCIe port. More than a couple NVMe drives is still in my opinion beyond an enthusiast class need because it sucks up available PCIe lanes so quickly.

It would be cool to see an enthusiast enclosure for say up to 8 drives back to a single PCIe attachment. The technology certainly exists. Microchip's PFX 28xG5 PCIe Fan-out is basically built for that sort of thing, but at over $300 per chip in small quantities I think it will be a long time before we ever see these built into something like Icy Docs. They're mostly relegated to server backplanes for All-NVMe Storage when the host doesn't have enough lanes in the design for direct-to-NVMe.
Yeah adding $300 to the cost of the encloser/fan-out is a nonstarter. Woof.
 

biostud

Lifer
Feb 27, 2003
18,846
5,707
136
E31T is not Gen 4 speeds (up to 10.8GB/s).
But the real solution is other new controllers made on more advanced manufacturing processes which Phison is too cheap to do yet.

Or perhaps to use conventional 2.5" sized SSDs that connect via U.2/U.3.
E31T is made on 7nm, while E26 is made on 12nm, so they have already started. I think we will see gen5 drives that doesn't require cooling by the end of 2024.
 
Reactions: Hotrod2go

BFG10K

Lifer
Aug 14, 2000
22,709
2,997
126
My 990 Pro 4TB rarely gets above 40c. Unless I am copying large files between nvmes.
Bare drive without heatsink easily throttles.


In fact you don't even need gen4, virtually every bare gen3 drive can temperature throttle just by copying large files to it.
 
Last edited:

jarablue

Member
May 3, 2004
120
20
81
Blah blah. I was told by multiple people that the 4080 was not a good 4k card either. Been doing this stuff so long that I only take what I experience as truth and not someone else saying this is how it is. I have a P3 Plus drive that I copy files back and forth between the 990 and it and both never throttled. Take what you hear someone say as fact with a grain of salt. Younger people do not like others challenging their opinion. Nature of being young.
 
Last edited:

CP5670

Diamond Member
Jun 24, 2004
5,564
645
126
From my PoV I just want something serviced with enough PCIe lanes that I don't have to worry about whether the PCIe slot my video card is in going down from 16x to 8x because I put in too many drives. That or if there aren't enough lanes in total for all the drives I've installed, I'd want them to be shared instead of stealing speed from the video card slot.
You really have to check the documentation for the specific board on this. It's hidden in the manual somewhere. Mine has the last pcie slot shared but the main video card slot always remains at 16x.

My 990 pro runs at more like 50-60C. I just have the board's passive heatsink on it, which is little more than a slab of metal, but it's good enough.
 

DavidC1

Golden Member
Dec 29, 2023
1,211
1,932
96
Moving to a new standard is NOT the solution, because then we'll truly have SSDs with heatsinks the size rivaling CPUs.

The real fix is stop increasing sequential speeds like it's a new fad. And that's what it is. Increasing numbers of no relevance.
I think the solution is to wait for 7nm controllers instead of 12nm.
You can't have the newest processes, because SSDs are low margin. N3 for example would break them. If they do though, you can be sure it'll be Gen 7 using even more power than 12nm ones.
 
Reactions: Igo69

mikeymikec

Lifer
May 19, 2011
18,982
12,103
136
Moving to a new standard is NOT the solution, because then we'll truly have SSDs with heatsinks the size rivaling CPUs.

There's a few posts here describing why it would be a good solution due to the disadvantages of the M.2 design.

The real fix is stop increasing sequential speeds like it's a new fad. And that's what it is. Increasing numbers of no relevance.

IMO that's an issue that market forces will likely take care of: While there are (and will always be) people who chase the latest thing as if their lives depend on it (and some obviously need that performance regardless of disadvantages), I for example see no personally applicable reason to go for gen5 SSDs as they currently are. I'm in the process of replacing my Linux SSD and only because my new case really isn't an ideal design for multiple SATA drives, so I'm switching my old and otherwise fine Samsung 840 PRO 256GB SATA SSD for a 970 Evo Plus 1TB (and I only went for that size because it was the same price as the 500GB model and no 250GB model was in stock at the time).

Furthermore, most people with experience of hardware with titchy fans are aware of their tendency to whine as well as failing far sooner than larger fans (likely due to the latter's lower rotational speeds to achieve the same/better airflow).
 
Reactions: CP5670

biostud

Lifer
Feb 27, 2003
18,846
5,707
136
You can't have the newest processes, because SSDs are low margin. N3 for example would break them. If they do though, you can be sure it'll be Gen 7 using even more power than 12nm ones.
Phison has literally just announced the E31T 7nm low powered controller. It's not going to be the fastest out there, but a first step.
 

BFG10K

Lifer
Aug 14, 2000
22,709
2,997
126
Been doing this stuff so long that I only take what I experience as truth and not someone else saying this is how it is.
There's literally objective proof in the video with the FLIR camera showing the temperatures. Also multiple reviewers show the drive throttles under sustained writes.

Younger people do not like others challenging their opinion. Nature of being young.
What's actually happening is some people develop an emotional attachment to their hardware, so any objective criticism of said hardware is perceived as an affront to their manhood.
 
Last edited:

BFG10K

Lifer
Aug 14, 2000
22,709
2,997
126
Phison has literally just announced the E31T 7nm low powered controller. It's not going to be the fastest out there, but a first step.
But virtually every bare gen4 and gen3 drive still throttles even after die shrinks, so you're still going to need a heatsink of some sort on gen5. It may not be as big as before, but it still won't fit into laptops or other small form-factors.

Actually some bare QLC drives don't throttle because their sequential writes already drop to ~130MB/sec, which is far below any modern HDD. Rather ironic actually.
 
Reactions: igor_kavinski

biostud

Lifer
Feb 27, 2003
18,846
5,707
136
But virtually every bare gen4 and gen3 drive still throttles even after die shrinks, so you're still going to need a heatsink of some sort on gen5. It may not be as big as before, but it still won't fit into laptops or other small form-factors.

Actually some bare QLC drives don't throttle because their sequential writes already drop to ~130MB/sec, which is far below any modern HDD. Rather ironic actually.
Is there any indication of throttling under gaming/desktop use or is it only when you throw continuous workloads on them?

As a casual gamer/desktop user is there any reason to worry about throttling?
 

BFG10K

Lifer
Aug 14, 2000
22,709
2,997
126
As a casual gamer/desktop user is there any reason to worry about throttling?
The vast majority of people won't notice difference between SATA and NVMe unless performing large file copying or another kind of niche I/O.

Bring on real SATA-4. A bump to 1GB/sec would be absolutely fine. Keep bumping the sequentials to stay far below manufacturing improvements and you'll never have a problem with runaway thermals or exotic form-factors.
 
Reactions: igor_kavinski

biostud

Lifer
Feb 27, 2003
18,846
5,707
136
The vast majority of people won't notice difference between SATA and NVMe unless performing large file copying or another kind of niche I/O.

Bring on real SATA-4. A bump to 1GB/sec would be absolutely fine. Keep bumping the sequentials to stay far below manufacturing improvements and you'll never have a problem with runaway thermals or exotic form-factors.
I know, but lets say texture streaming/RTX IO etc. become a thing, will it actually load the drive enough to throttle?

If you install a gen5 drive with a normal heatspreader and well ventilated case, will it even be a problem?
 

DavidC1

Golden Member
Dec 29, 2023
1,211
1,932
96
I know, but lets say texture streaming/RTX IO etc. become a thing, will it actually load the drive enough to throttle?
Only if it's a significant contribution.

To take full advantage of sequential speeds, games have to be architected in a way so it'll benefit from whatever an SSD needs to max it out. Since that seems to be 8+ threads at 1MB working size, I doubt it'll get anywhere near there even if it's well optimized.

Sequentials for loading speeds is basically a fantasy. Actually considering even most file transfers aren't pure sequentials with large files, even that's a fantasy.

Optane would have been a boon if they continued it. Except new technologies like that suffer from chicken and egg problem regarding volume and cost. Volume is low, so costs are high, so volume is low so costs and high and on and on.

One analyst said that Optane needed to get within 10x of the volume of DRAM to realize it's full potential, while making the inventor money. Leakers have said that Intel barely made money on the 905P SSD that cost $1000 for 480GB so that should give you a clue.

True, I think their strategy could have been better. But with Gelsinger at the helm, any memory-related ventures would have died. He was mentored under Andy Grove, who's famous for turning Intel around by changing it from a memory(DRAM) centric business to a MPU focused one.

Pat "I never want to be in memory" Gelsinger.
 
Last edited:

DavidC1

Golden Member
Dec 29, 2023
1,211
1,932
96
The vast majority of people won't notice difference between SATA and NVMe unless performing large file copying or another kind of niche I/O.

Bring on real SATA-4. A bump to 1GB/sec would be absolutely fine. Keep bumping the sequentials to stay far below manufacturing improvements and you'll never have a problem with runaway thermals or exotic form-factors.
I think NVMe is faster by 10-20us than SATA. SATA SSDs get 4K Random read latency at 120-150us while NVMe can go little under 100us. Intel created the NVMe spec just so Optane on NVMe didn't look even more terrible than it did.

NVMe spec but with SATA physical connectors is maybe what you're looking for?
 

biostud

Lifer
Feb 27, 2003
18,846
5,707
136
A review from TPU showing it doesn't require active cooling, and first starts throttling after 3TB continuously writing.
 
Reactions: igor_kavinski

AdamK47

Lifer
Oct 9, 1999
15,635
3,410
136
Trigger warning...
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.


MSI SPATIUM M580 PCIe 5.0 NVMe M.2 FROZR LIQUID SSD

 
sale-70-410-exam    | Exam-200-125-pdf    | we-sale-70-410-exam    | hot-sale-70-410-exam    | Latest-exam-700-603-Dumps    | Dumps-98-363-exams-date    | Certs-200-125-date    | Dumps-300-075-exams-date    | hot-sale-book-C8010-726-book    | Hot-Sale-200-310-Exam    | Exam-Description-200-310-dumps?    | hot-sale-book-200-125-book    | Latest-Updated-300-209-Exam    | Dumps-210-260-exams-date    | Download-200-125-Exam-PDF    | Exam-Description-300-101-dumps    | Certs-300-101-date    | Hot-Sale-300-075-Exam    | Latest-exam-200-125-Dumps    | Exam-Description-200-125-dumps    | Latest-Updated-300-075-Exam    | hot-sale-book-210-260-book    | Dumps-200-901-exams-date    | Certs-200-901-date    | Latest-exam-1Z0-062-Dumps    | Hot-Sale-1Z0-062-Exam    | Certs-CSSLP-date    | 100%-Pass-70-383-Exams    | Latest-JN0-360-real-exam-questions    | 100%-Pass-4A0-100-Real-Exam-Questions    | Dumps-300-135-exams-date    | Passed-200-105-Tech-Exams    | Latest-Updated-200-310-Exam    | Download-300-070-Exam-PDF    | Hot-Sale-JN0-360-Exam    | 100%-Pass-JN0-360-Exams    | 100%-Pass-JN0-360-Real-Exam-Questions    | Dumps-JN0-360-exams-date    | Exam-Description-1Z0-876-dumps    | Latest-exam-1Z0-876-Dumps    | Dumps-HPE0-Y53-exams-date    | 2017-Latest-HPE0-Y53-Exam    | 100%-Pass-HPE0-Y53-Real-Exam-Questions    | Pass-4A0-100-Exam    | Latest-4A0-100-Questions    | Dumps-98-365-exams-date    | 2017-Latest-98-365-Exam    | 100%-Pass-VCS-254-Exams    | 2017-Latest-VCS-273-Exam    | Dumps-200-355-exams-date    | 2017-Latest-300-320-Exam    | Pass-300-101-Exam    | 100%-Pass-300-115-Exams    |
http://www.portvapes.co.uk/    | http://www.portvapes.co.uk/    |