SSD Benchmarks Hierarchy: We've tested over 100 different SSDs over the past few years, and here's how they stack up.

Soaptrail

Distinguished
Jan 12, 2015
303
96
19,420
Hard to tell Toms is a tech site when they restrict the table width to something so narrow you cannot see all the columns. I guess no one has embraced 16x9 monitors who work there, or do they only work on their phones in portrait mode?
 
Hard to tell Toms is a tech site when they restrict the table width to something so narrow you cannot see all the columns. I guess no one has embraced 16x9 monitors who work there, or do they only work on their phones in portrait mode?
Unfortunately, it's not a Tom's Hardware decision but a Future decision. But we did set the "wide" layout after the initial publishing, so give it a look now and it should be much better.
idle, average daily work, fully loaded temperatures and power consumption (with and without a heatsink) would be really good to know
I've just added a "Specifications" column that links to the appropriate review — there are a few drives that haven't been reviewed yet (pending), but everything else should have most of the details you want.

Temperatures aren't something we've really delved into, though I've collected the data on all of the more recent tests. The TLDR is that any motherboard with a decent heatsink, or a case setup where there's a fan blowing air over the SSD, should be fine. If you have a Gen5 drive without a HS, or covered by a hot graphics card and in a case with restricted airflow, yeah, they can get hot and throttle. But then is that an SSD problem or a system problem? I generally point to the latter, and it's why I don't much care for mini-ITX builds personally.
 

dimar

Distinguished
Mar 30, 2009
1,103
95
19,360
Temperatures aren't something we've really delved into, though I've collected the data on all of the more recent tests. The TLDR is that any motherboard with a decent heatsink, or a case setup where there's a fan blowing air over the SSD, should be fine. If you have a Gen5 drive without a HS, or covered by a hot graphics card and in a case with restricted airflow, yeah, they can get hot and throttle. But then is that an SSD problem or a system problem? I generally point to the latter, and it's why I don't much care for mini-ITX builds personally.
I was asking because I'm having trouble finding the perfect SSD for a laptop that doesn't destroy the battery or that doesn't create lots of heat.
 
I was asking because I'm having trouble finding the perfect SSD for a laptop that doesn't destroy the battery or that doesn't create lots of heat.
If you want large capacity, the best bet is the Samsung 990 Pro 4TB. Alternatively, the Addlink A93 and Lexar NM790 seem like decent options, though obviously those are less reputable brands than Samsung — not sure I'd want to go that route just to save $40 or whatever. I'd generally go with the 990 Pro 2TB as a very good laptop pick. It can get a bit warm under sustained loads, but those aren't normally a real problem with SOHO use.
 

Soaptrail

Distinguished
Jan 12, 2015
303
96
19,420
Unfortunately, it's not a Tom's Hardware decision but a Future decision. But we did set the "wide" layout after the initial publishing, so give it a look now and it should be much better.

I've just added a "Specifications" column that links to the appropriate review — there are a few drives that haven't been reviewed yet (pending), but everything else should have most of the details you want.

Temperatures aren't something we've really delved into, though I've collected the data on all of the more recent tests. The TLDR is that any motherboard with a decent heatsink, or a case setup where there's a fan blowing air over the SSD, should be fine. If you have a Gen5 drive without a HS, or covered by a hot graphics card and in a case with restricted airflow, yeah, they can get hot and throttle. But then is that an SSD problem or a system problem? I generally point to the latter, and it's why I don't much care for mini-ITX builds personally.
Thank you. I figured it was not your fault and instead a decision the higher ups with no skin in the game that frustrates you and your colleagues as much as us readers.
 

saunupe1911

Distinguished
Apr 17, 2016
212
76
18,660
I only trust Samsung SSDs.

I learned my lesson the hard way. I got two unopened replacement Sabrent 4 NVME that will never get used because four of them died on me since 2020. But hey Sabrent will just keep replacing the SOBs smh.

And dang you guys haven't tested a 1 TB 990 Pro??? It's not even charted
 
  • Like
Reactions: Order 66

tyns78

Honorable
Oct 25, 2017
13
2
10,515
It's a shame you dropped the p5800x from these comparisons (and even the 900p/905p). I understand they're not made anymore and few of your readers will ever understand why they're so special let alone be willing to spend the big money on buying one but an SSD this site proclaimed as "the fastest SSD ever" getting swept under the rug does a disservice to all of your "best of the best" and ranking posts makes them feel disingenuous, turning it into glorified clickbait. Show us when the tech that killed 3DXP actually comes close to its performance in real world usage.
 
It's a shame you dropped the p5800x from these comparisons (and even the 900p/905p). I understand they're not made anymore and few of your readers will ever understand why they're so special let alone be willing to spend the big money on buying one but an SSD this site proclaimed as "the fastest SSD ever" getting swept under the rug does a disservice to all of your "best of the best" and ranking posts makes them feel disingenuous, turning it into glorified clickbait. Show us when the tech that killed 3DXP actually comes close to its performance in real world usage.
I do still have the data for it, and the random IO at QD1 was of course exceptional. RIP, Optane. RIP. For those that are wondering what the fuss was about:

1700528713341.png
 

Heiro78

Prominent
Nov 20, 2023
57
36
560
You list the 4TB TeamGroup A440 Pro but I believe you tested the A440 Pro Special Series. I don't believe the non-special series performs the same as the special series. Can you confirm since there is an about 30 or 40 USD difference?
 

bit_user

Titan
Ambassador
Hard to tell Toms is a tech site when they restrict the table width to something so narrow you cannot see all the columns.
Try reducing the width of the window. At some point, the page will switch to single-column layout and you'll get more width for content (including tables). It's still not enough for some of the tables, but the ones in this article display for me without any horizontal scrollbars.

the Graph is very hard to read without heavy zoom.
Click the black icon, in the lower-right, to get a higher-res version of the image. For me, this is enough zoom if I make my window sufficiently wide.
 
Last edited:
  • Like
Reactions: Order 66

bit_user

Titan
Ambassador
This year, I bought a 3.84 TB Solidigm (formerly Intel) D7-P5520:


I'm still buying parts for the machine it's going into, so I have no personal observations or data to add about its performance on client workloads. Anyway, I don't game so I can't comment on how well it'd work for that. I'd be curious to know how it would handle something like Starcraft, with its industry-leading tail-latencies!

I bought mine here:


One downside is that most consumer motherboards will require an adapter of one sort or another. However, the biggest downside is probably its high idle power of 5 W. That's pretty typical of enterprise SSDs, it seems, but much higher than consumer SSDs. Because of that, I think you really ought to mount it directly behind a front intake fan, rather than on one of those PCIe AIC host cards.

It's a shame you dropped the p5800x from these comparisons (and even the 900p/905p). I understand they're not made anymore
You can still buy them. Believe it or not, this is about half what they cost, compared to 1 year ago.


I would still recommend the D7-P5520, though. Roughly 10x the capacity for about half the cost, and you still get good enough QD1 and tail latencies low enough that most people will never notice the difference. Lower power consumption, too.
 
Last edited:
  • Like
Reactions: JarredWaltonGPU

bit_user

Titan
Ambassador
Temperatures aren't something we've really delved into, though I've collected the data on all of the more recent tests. The TLDR is that any motherboard with a decent heatsink, or a case setup where there's a fan blowing air over the SSD, should be fine.
How do you mount SSDs on motherboards with their own integrated heatsink for the slot? Do you use a thermal pad, a blob of heatsink compound, or just rely on there being enough contact for adequate heat conduction?
 

bit_user

Titan
Ambassador
What I'd like to see is a list of SSDs ranked by performance/watt efficiency, while still delivering good performance. It would be very useful to squeeze more battery out of gaming notebooks and portables, like the Steam Deck.
IMO, this sounds more useful than it is, in practice. PCIe 4.0 and 5.0 drives tend to win on perf/W, because they generally increased performance beyond the amount of additional power consumption. Does that mean PCIe 5.0 drives have any place in a notebook? I'd say not.

What I'd look at is how much performance SSDs provide that stay within a certain active power budget - maybe one range for gaming laptops and another for ultrabooks. The other thing to look at is their ranking by idle power.
 

bit_user

Titan
Ambassador
I was asking because I'm having trouble finding the perfect SSD for a laptop that doesn't destroy the battery or that doesn't create lots of heat.
SK Hynix P31 Gold is a good choice, if you don't need top performance. Perhaps it no longer has the best idle power, however.

If you want more performance than that, search the article for drives with a "Phison E21T" controller. Those have slightly higher average active power, but a little better performance on all fronts. It's unclear how they'll perform in a PCIe 3.0 slot, but they might still edge out the P31 Gold while using closer to its level of power consumption.

"Unprecedented efficiency"

But the original review is more than 3 years old. Look at the list of updates: it was first published in Sept, 2020.
 
Last edited: