That’s totally OK. See you end of April, when we have our shiny new devices to show off.
I certainly hope you do!
Assuming there isn’t another delay…
[quote=“hosh0, post:193, topic:5705, full:true”]
Can I just ask that those who prefer performance not be forgotten about? Battery life can be replenished by a charge. But once you sacrifice that performance, that’s it you can’t do anything to get it back.[/quote]
Here’s the thing about SSD performance; on this type of devices with the usage pattern, PCIe SSD provides little to no benefit as the rest of the hardware isnt that fast either. Remember, we are not building an X99 workstation or a server. Meanwhile, 0.3W less power on average means one extra hour of battery life.
Check out the discussion here https://eve.community/t/ssd-speeds-how-large-is-the-difference-between-sata-and-pcie/1783 and here https://eve.community/t/how-fast-can-we-go-speed-of-the-ssd/1667
That’s your opinion that it doesn’t offer much benefit. For my use case it certainly does, I’m not an average user and I understand that I’m probably an outlier here. But for my use case where I do lots of facial matching prototyping and the prototypes need to access millions of small image files, for me it’s extremely useful. Given I work in an office environment 99% of the time, an additional battery time of one hour means nothing.
Now I’m not saying I’m right or anything, but rather showing that people need to consider use cases that aren’t their own. Everyone’s use case is just as valid as the next persons and Eve as a company have to make decisions on what they feel is the right balance for the V.
If it was up to me, I’d love to see the option for a Samsung Pro based SSD. Again however small the performance difference to you, it’s a difference that is lost and can’t be made up. Batteries can be recharged.
I doubt the Core M we’re using is suitable for your kind of work anyway…
However what is the benefit of high perfomance if my battery is empty ?
Thats not necessarily better in a PCIe SSD. Its better to know which SSD exactly we are using, instead of judging the performance just from the interface. While I agree that PCIe gives you the bandwidth to operate at much faster speed than SATA, not all drives take advantage of it. The best example is, once again, the Intel 600p.
As you can see here, the 600p (PCIe) is actually slower in reading small files than 850 EVO (SATA), up to 42% slower in fact.
Worse, it does all that while consuming more than twice (!) as much power
As I stated before, PCIe is not a guarantee for fast performance. It is down to each individual SSD itself.
Take a look at my post above, while Pro series might be too expensive for the V, Id say that the PM961 (similar to 960 EVO) has 90% of the performance while being not-that-much more expensive than the 600p, and it is more power efficient at the same time. The Plextor M8Pe is also one of those fast and power efficient PCIe SSD. Worst case scenario, even the 850 EVO is still faster than the 600p
Fully agree pcie doesn’t just make it faster. Which is why I wanted to see benchmarks from the 600p in the V. If Eve have been able to get into engineers to increase the performance of the drive in relation to the V , I’d want to see that. Mike clearly stated Intel is giving them extra support and access to their engineers so I just want to see the outcome of that before deciding.
EVE next project and why it should be a laptop (update)
While the benchmark from Anandtech is totally valid,
it should be noted that the test focuses on sustained performance but not burst performance.
Real world use cases may result in differing experience.
I would be happier to just have a 850 EVO which is consistent, more power efficient and worse “best-case scenario” performance.
I seriously doubt some engineering optimization could make up for 42% lack of speed. The test above is done in a desktop PC, a platform where performance takes the front seat, and power usage is pretty much in the lowest priority. Optimizing power usage and curve is one thing, but increasing speed from engineering integration is very unlikely, in my opinion. Im not saying its impossible, but please do not be surprised if the speed outcome is not any faster than the table above, especially after the system is tuned for mobility (power saving).
Worse, if the 600p SSD is already paid, then youre pretty much stuck with slower-than-SATA SSD performance that consumes up to twice as much power. Thats why IMO it is important for us to decide as soon as possible, and tbh the 600p is not really impressing me from any point of view. What do you think?
Totally agree. I am doubtful that Intel could change the characteristics by V firmware updates to make up the difference in speed. Several of the options that @Patrick_Hermawan mentioned have better performance and better power out-of-the-box over Intel.
The Intel solution was designed to be a mid/low performer using the PCIe interface. If we have time, we can look at the benchmarks from the prototype team, but like others have mentioned some of the other choices show a much better performance-energy consumption that is hard to beat. Just my 2 cents.
Not if you’re in the middle of Manhattan! What any of these tablets (all of which suffer fro poor battery life) really need is a pop out battery, like I have in my Cygolight bicycle light, or even like my cell phone. Carry a spare, you know? Why can’t anybody to that?
Given the battery is built-in, just take a powerbank with you if you need more capacity - you can use that to charge other devices too.
Because you couldn’t squeeze in the same size battery that way. Now it’s fit in every little space there is left in the device. With pop-out you’d have to have a “standard” form factor (e.g. rectangular) to make it easy to use. Plus you’d go away from the unibody design…
I second that. Battery is the most important factor. And by the way: If we have a SSD with ~500MB/s that is still fast enough if it means more battery life. Because honestly I do not care too much if write speed is 1GB/s or 500MB/s in the V because I won’t have huge amounts of data that need to be written sooooo fast because
- as most SSD reviews show: reading speed has no big impact on normal every day performance (program launching times are the same, or who would desperately need 5.8s launch time instead of 5.9s?)
- and write speed on 600p is no better than 850 Evo for example.
- copying data from/to V is limited by Gigabit networt speed anyway.
- copying/moving data on the V or extracting archives will never be sooo important (at least for me…) that I would care. Most of the times those tasks are one-time things that are not time critical where I can wait some more seconds.
TLDR: I would definitely prefer a SSD with the lowest possible power consumption.
Should we do a poll? Should I make a separate thread for the SSD discussion?
@Patrick_Hermawan - yes, let’s do a poll. I think we have the opportunity to confirm what is important for the group. Originally it was battery life (especially since the current prototypes are obtaining 7-8 hrs, having an efficient ssd that adds an 1+ hrs (without further firmware optimization) would greatly assist in guaranteeing the 10-12 hrs we were targeting.
Before actually voting, I would like to know from @Mike if this is all too late, that the first Vs would use the 600p anyway.
Otherwise I believe most of the audience of this thread is probably well-informed about SSDs enough for an unbiased vote.
good point. I asked @Mike what parameters/options are available and if this was something that could be implemented.
I understand they are quite busy, hopefully he has a chance to confirm what parameters are stuck (e.g. 2280 form factor) and what parameters can be changed (including which batch these changes could be implemented on).