Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Intel G3 SSD Delayed
- Thread starter Cynicalone
- Start date
- Sort by reaction score
Thanks for the heads up! Hopefully Intel learned from G2 that they shouldn't launch something that they cannot deliver. It doesn't matter is it launched in November 2010 or February 2011 if it doesn't hit the shelves before February 2011.
Yeah, I think they are probably being much more cautious this time around in terms of solidifying firmware and controller operation before launch. G3 drives will need much more than 25nm NAND to be competitive. They are undoubtedly redesigning their controller and looking at compression techniques like Sandforce to bring down the write amplification to maximize performance... which translates into significant validation testing for this gen.
Well, the Sandforce 2000 chip will give the G3 a beating.
G3:
Sequential Read: 250MB/s
Sequential Write:170MB/s
read IOPS: 50K
write IOPS:40K
Security: AES-128
Sandforce 2000:
Sequential Read: 500MB/s
Sequential Write:500MB/s
read IOPS: 60K
write IOPS:60K
Security: AES-256
Source: http://www.anandtech.com/show/3971/...troller-sf2000-capable-of-500mbs-and-60k-iops
G3:
Sequential Read: 250MB/s
Sequential Write:170MB/s
read IOPS: 50K
write IOPS:40K
Security: AES-128
Sandforce 2000:
Sequential Read: 500MB/s
Sequential Write:500MB/s
read IOPS: 60K
write IOPS:60K
Security: AES-256
Source: http://www.anandtech.com/show/3971/...troller-sf2000-capable-of-500mbs-and-60k-iops
Actually, it's 375MB/s. It's 3 Gigabits per second and 8 bits are equal to 1 byte. Thus you divide 3Gb/s (3000Mb/s) by 8 and you get 0.375GB/s = 375MB/s.
However, the real world max is ~285MB/s
He was right that the unit used, Mb/s (after he converted it from Gb/s), needs to be divided by 8 to get MB/s. So the theoretical maximum is 375MB/s.have a feeling there's 2 bit's of overhead in there so you need to divide by 10. dont shoot me if i've got it wrong tho
But SATA does use an 8b/10b encoding scheme, and a user will never see the theoretical maximum in practice (due to latency).
You said everything, but you did not explain it 100% correctly. Latency is not an issue in this case. SATA II is capable of transferring 300'000'000 bits per second, but every payload byte is coded as a 10bit-sequence (for redundancy, clock recovery and DC-balance reasons)
The theoretical maximum of the physical link is therefore 300MB/s payload, which is somewhat reduced due to protocol overheads. Which leads to the mentioned ~280MB/s real-world performance.
I was trying to keep it as simple as possible.
The 375MB/s figure is a max throughput, but it's the burst rate, not sustained. I don't see a need to get into clock recovery (PLL circuits), DC coupling,...
The figure you've posted is the maximum uncoded transfer rate, which is 2.4 Gbit/s = 300 MB/s, which is essentially the maximum sustained throughput. From this figure to real world results, the throughput loss is much lower (~20 - 30MB/s) due to the protocol (primarily), but it still meets the definition of latency (latency = time delays experienced in a system, regardless of the source). Part of this loss may also be from other issues as well, such as jitter.
maybe there's still hope for us yet!
in the past has intel made announcements about delays?
Unless you are buying a new Mac with an updated SATA controller, there is no point in waiting.
The G3 will still be SATA II, but with heavily increased random performance, better sequential writes and a better price/GB rate.
So there's definitely a point in waiting.
I sold of all my ssd's to wait for this one. It will have a 600gb size for the max one. That size will hold all my go to info ie a 400gb itunes and around 100gb of programs etc. having my entire itunes on an ssd (not 2 in raid0) has been a goal for me.
According to AnandTech, there may be models with SATA 6Gb/s as well
http://www.anandtech.com/show/3965/intels-3rd-generation-x25m-ssd-specs-revealed