Adata SX8000 NVMe SSD Is The First With Intel 3D MLC

  • Thread starter Thread starter Guest
  • Start date Start date
Toggle sidebar Toggle sidebar
Status
Not open for further replies.
Right now the race is for speed among the NVMe drives.
When the speed is about the same along all the drives manufactures will start dressing up the drives with fancy colour heatsinks.
Then when those get stale they will add RGB LEDs.
 
I agree with this, except I think we'll see crazy heatspreaders and LEDs sooner than that. It'll be the drives which can compete on neither price nor speed.
 
2000/600 Mb/s and 80k/140k IOps for 256/512 Gb versions, really? When there are 960 EVO and 960 Pro with 3500/1500 Mb/s and 330k/300k IOps?

This thing should be really cheap. But even in that case there is no point to choose it over 130$-priced 960 EVO.
 
Why no u.2 format? The m.2 setups are a pain to do on a desktop, and 2.5 mounting is much easier. (those m.2 2mm screws are a very very small)
 
I understand that the switch to NVMe was necessary to move forward with the technology.. but it makes no sense when we're discussing performance and NVMe is touted as a class of it's own. This isn't the only article I've seen doing this, but they often seems to read as "NVMe vs low performance". But in practice, NVMe didn't bring a huge performance increase itself.
"true NVMe performance", and "NVMe starts to outsell SATA". NVMe is an apple and SATA is an an orange.. or more appropriately a stem.

This should be "SATA vs PCIe 2.0 x4" or "NVMe vs AHCI".

http://www.anandtech.com/show/9396/samsung-sm951-nvme-256gb-pcie-ssd-review
 
There are several PCIe AHCI products like the Samsung SM951-AHCI and older All-In-One RAID products.

In the future we will combine SATA and NVMe in one large group but there are not enough low-cost NVMe products to make the merge just yet. They are coming, the Adata SX8000 may be one of the NVMe SSDs priced at premium SATA levels but at this time we don't have pricing info.

As far as U.2 in the client space, I think it is dead. It will go down as only mildly more successful than SATA Express. If Intel had a M.2 controller a year ago then I don't think we would have ever had the U.2 spec on consumer motherboards. It was designed for enterprise rackmount systems with dedicated drive sleds. It works great for that purpose but in consumer systems the cable is out of place, and companies have to cool hardware the same hardware with a smaller footprint. That costs more money and requires more cooling from the system.

That is not to say we will never see a new consumer-facing U.2 SSD. Personally I like the adapters and the ability to utilize enterprise SSDs on consumer boards. SCSI and SAS controllers were always expensive and usually suffered from compatability issues with some motherboards.
 


It actually means a hell of alot... we don't need to have the fastest as the difference just isn't here, and with a good number of games, we just need a properly defragged hard drive. Still considering getting a 512gb ssd for games that i play allot or the odd fallout 4, but my 4tb seems to do the job just fine.
 


That was done with a stopwatch, and he didn't even ensure prefetch was disabled/enabled. It was done so poorly you can dismiss it outright. That said, load times don't matter. The improvements from an NVME SSD are not going to be easily measured but felt and measurable with a proper benchmarking (which is not easy, which is why few to none are doing it). If you're feeding a lot of texture information off the drive and other things are also requesting the drive, you'll see a significant difference in how the game runs.
There's a lot of bad or imperfect benchmarks out there, but faster is always better. Buy the best you can afford because every bit counts as it ages. Anyone buying a SSD today should at least get an NVME model. There's no reason not to.
 


No reason for heatsinks when an SSD has a power consumption of 2W.
 


yes and no. I remember watching the screen savers when quake 3 was new, they showed off a standard pre built pc load the level, then they showed the difference that a 10k rpm or 15krpm drive made on the same level. my current 4tb drive kicks that absolute crap out of anything that was available back then outside of a ram drive.

There is such a thing as diminishing returns, and as far as ssds go, we hit that with sata 2 drives, much less sata 3.

sure faster is always better, however at a price premium when the use cases where we notice that difference does not justify the cost outright. Hell, I have an extensive image archive, to open some of my folders off a hdd could take up to 10 minutes. on my desktop i have a FAR smaller folder like this that is 1556 images big 1.29gb, but it takes 26 seconds to open off an ssd, i have a second on that is slightly larger that at 5500 and 4gb, but this one loads almost instantly.
my ssd is an intel that was 200mb read, but has gone down to 100-120mb read over the years, I have no idea why that folder takes forever to load, but this is a case use where a faster drive would be better, and would like to transfer the entirety of the image archive off of hdd to ssd but I have a VERY hard time justifying the cost, much less in a gaming application where the gains are almost non applicable outside of a few VERY poor pc port or horrific engines.
 
I don't know what the peak dissipation of the faster NVMe drives is, but you should know that they do throttle performance due to overheating, in real-world scenarios.
 


It's not going to overheat.
 
Seriously, stop making things up. This is not the first time. If you don't know what you're talking about, at least do a quick web search, instead of spreading false information. And doubling down only makes it worse.

Some search results for "ssd thermal throttling":
 

Please calm down. 😛 Perhaps instead of heatsinks they should look at reducing the concentration of wasted energy. That, or using some form of material with better heat tolerance. There are other approaches besides heatsinks.
 


There really isn't a plethora of different materials to make the chips with to improve thermal tolerances and you can't do much about the concentration of waste heat when you're working with these chips in a tiny form factor. The fact is that as we make things faster and put them in such a small form factor, heat starts to build up and it needs to go somewhere. What we've been doing is constantly improving on the efficiency of the components, especially through smaller process nodes and better controllers, but that's only taking us so far right now.

The biggest problem is probably the lack of room for heatsinks on M.2 drives since many of them are meant for laptop use where there isn't always enough room for chips to be on both sides of the M.2 device, let alone anything more. You might not be able to get anything more than a paper thin aluminum sticker over the chips in some computers.
 


make the damn things thicker? that is my solution to not having proper heat sync, also how to fit more batteries in the thing too.
 
But most people want thin and light, and the standards need to follow the market.

A friend just bought a laptop/tablet, and he required that it have no fan. He was even willing to sacrifice performance for that.
 


Intel chips seem to be fine up to 100C whereas AMD chips struggle around 70C. Apparently there is some difference in materials. I don't know how that translates into SSDs. But if overheating M.2 SSDs were a real problem would not manufacturers like Samsung already have addressed the issue? Apparently in the real world it's not a problem at all. I find it hard to believe this could be a big issue pointed out by some people on forums, that SSDs need heat sinks, where the entire SSD faculty at Samsung would not have considered this and thought about it thoroughly.
 
I'm no semiconductor expert, but I think it's a mistake to compare SSDs with CPUs. SSDs are trying to reliably and persistently store about 100,000 times as much information as a typical CPU holds (if you add up the capacity of a typical CPU's cache hierarchy and registers, it's on the order of 10 MB). More heat means more entropy, which is the enemy of persistent information storage. Hopefully, someone more knowledgeable can chime in on the specifics, or else maybe you can find it written about on the web.

Paul recently wrote about carbon nanotube storage that uses electrostatic charge and is far more heat resistant. The only problem is that it's also lower density.

Well, it depends on how much ventilation you have, for one thing. And it takes a while for the heat to build up. So, it would be a subset of cases where this would become noticeable. It certainly does happen, and can be an issue for some people.

Samsung's 960 announcement made a pretty big deal about the copper-infused product label doing double-duty as a heat spreader. Do you actually read the news, or just comment on it?
; )

If the race to make SSDs more dense were somehow put on hold and the focus turned exclusively to sustained performance, then they might be able to make more headway on the heat tolerance front. That's obviously not going to happen.

IMO, the real problem is that M.2 is just not a great form factor for high-end desktops. A 70 C thermal ceiling shouldn't be an issue, if cooling weren't so highly constrained.
 


did he require it to have no fan or did he want it to have no fan?
I honestly believe most people don't want it thin, and would rather have function and performance, however apple marketed everything as thin, and everyone tries to chase apple.



If i remember right, its a difference in the way intel and amd report temperatures, not so much different materials.

-- small edit for wrong company said --
 


Storage isn't really my Kia Forte. Looking at this article, everything flies over my head. " IMFT's new 256Gbit 3D MLC NAND." It's like robot code or something. There's NAND, 3D NAND, MLC NAND, NVME, M.2, now all this weird crap getting integrated into motherboards. It's all just coming too quickly for me to take it in. What about normal 2.5" SSDs? Have those been forgotten? Is SATA speed maxxed out? Why not just make SATA faster? I think I'll stick with power supplies where I'm good at. Which is why I posted here, I know the power requirements of an SSD is extremely small, and since that electromagnetic energy is not being converted into mechanical energy (as in a case fan), it's clearly all going to heat. Seeing as 20W GPUs can be cooled passively, I figured with a 2W SSD it would be a breeze (pun intended) to cool it.
 
In this article, Samsung predicts 2017 as the year PCIe will overtake SATA. Note that PCIe technically includes M.2, U.2, SATAe, and good 'ol add-in cards.

http://www.tomshardware.com/news/samsung-ssd-hdd-sata-nvme,32762.html

The evolution of SATA, as we have known it, is officially done. SATAe exists as a sort of transitional standard that I think still has legs but some have already written off:

https://en.wikipedia.org/wiki/SATA_Express

That SAS has gone another step (to 12 Gbps) shows that SATA could've kept going. Even then, it couldn't match the bandwidth of PCIe-based alternatives, nor the low latency of NVMe.

The real problem is all to do with M.2's mobile-oriented origins, forcing it to be too thin to accommodate a proper heatsink. Furthermore, the small board area results in a large number of chips (some of which have up to 768 layers of cells) being packed very densely. Couple that with a desktop case market that hasn't evolved to focus airflow on motherboard-mounted M.2 slots and you have a recipe for thermal throttling. Not saying it's a huge problem, but it's there.
 
Status
Not open for further replies.

TRENDING THREADS

Latest posts

Moderators online

Share this page

COMPANY
RESOURCES
FOLLOW
Top Bottom