A-DATA 32 GB 633x Compact Flash

[Darksaber]

The 633x Compact Flash from A-DATA is their fastest CF offering available today. With advertised read and write speeds of above 90 MB/s it should be an interesting performer. We take a close look and push it to the real world limits - for your reading pleasure.

Show full review

 

[MadMan007]

From page 3:

Hard drives tend to drop in speed the closer the reading head gets to the outer edge of the platters

Is this correct, or am I confused? If it's in regard to STR I thought that the outside edge of a platter is fastest because at a constant RPM for the whole platter there is more disc area=more data moving under the heads in a given amount of time at the outer edge. Think of a platter divided up like a pie chart, for any 'slice size' there is more area moving under the heads in a given amount of time at the outer edge than at the inner edge.

 

[Darksaber]

From page 3:



Is this correct, or am I confused? If it's in regard to STR I thought that the outside edge of a platter is fastest because at a constant RPM for the whole platter there is more disc area=more data moving under the heads in a given amount of time at the outer edge. Think of a platter divided up like a pie chart, for any 'slice size' there is more area moving under the heads in a given amount of time at the outer edge than at the inner edge.

Actually, the innermost area of the platter is the fastest. A drive spins at 7200 RPM for example, and when the read head is all the way on the inner edge, reaching data somewhere in that area is fastest, as the circumference is smaller there. Thus the wait is shorter until the data moves under the read head.

Now move the read head all the way to the outer edge of the platter. Here the read head needs to wait much longer until the data comes around to be read. Larger circumference = longer wait until the data makes it full circle.

 

[crazyeyesreaper]

its kind of like a CD or DVD if u look at one or the other after burning a file u will see it starts at the inside edge and works outwards same with the HDD platters as Darksaber said it far faster closer to the center then the other edge but i wouldnt say standard HDD is going to be all that slow at the outer edge raw speed is nice but whats most important is seek times etc as imagine if an old hdd can read 50mb/s minimum theres not really anything out there to saturate it its the seek time that kills it and thats what makes an SSD great anyway more on topic

HDDs CDs DVDs and Blu ray all read from the inner edge spiraling outwards like an old vinyl record

 

[MadMan007]

Actually, the innermost area of the platter is the fastest. A drive spins at 7200 RPM for example, and when the read head is all the way on the inner edge, reaching data somewhere in that area is fastest, as the circumference is smaller there. Thus the wait is shorter until the data moves under the read head.

Now move the read head all the way to the outer edge of the platter. Here the read head needs to wait much longer until the data comes around to be read. Larger circumference = longer wait until the data makes it full circle.

You are talking about rotational latency or seek time, I was talking 'in regard to STR' Actually I've confused myself a bit now because iirc the lowest seek time parts of a drive are also the ones with the highest STR, but those two things really shouldn't go together. Or do simple low-level benchmarkls like HDTune and HDtach chart them for mismatching parts of the drive?

crazyeyesreaper - Writing optical disks in particular is a whole other story, there are numerous different write strategies that drives use. No modern high speed optical drives write at the same RPM across the whole disk. Reading from an audio CD, record, or such time-coherent streaming data set works by the read head moving tracks more often at the smaller diameter section. Or to put it another way, 1 inch measured from center to edge of the inner part might hold as much data as 1/4 inch of the outer area (it really should be 1/pi I think) so the head has to move through more grooves or tracks in the inner section to extract the same amount of data.

Now you wrote a bit of a runon there heh But it seems like you're saying that seek time is dominant in drive performance, which is fine if we mean 'overall drive performance using real programs' because the OS doesn't write data as STR not to mention need to access it, but think about it - does that still apply to much more synthetic STR data patterns?

 

[Darksaber]

You are talking about latency or seek time, I was talking 'in regard to STR'

Ah yes, sorry I was talking bout seek time, not sequential transfer rates

Well hard drive are set to fill data from the outer edges to the inner edges. So yes, STR would be the exact opposite of seek time, meaning that the outer edge would be better performing than the inner edge - if I remember correctly.

Added that to the review to avoid confusion.

 

[MadMan007]

No problem, it's really a minor thing as far as the overall review I was really just geeking out

This is what I mean by HDTune in this case appearing to match up the highest STR with the best (lowest) seek time:

Simple benchmark programs like this could certainly confuse people if they think it means the same part of the drive gets the highest STR AND the best (lowest) random seek time. So does this mean that for HDTune in this case the STR and seek time data are kind of misrepresented and should be taken separately rather than read as being on the same part of the drive?

 

[Darksaber]

No problem, it's really a minor thing as far as the overall review I was really just geeking out

This is what I mean by HDTune in this case appearing to match up the highest STR with the best (lowest) seek time: http://i114.photobucket.com/albums/...Benches/HDTune_Benchmark_ST3500630AS_3AAE.jpg

Simple benchmark programs like this could certainly confuse people if they think it means the same part of the drive gets the highest STR AND the best (lowest) random seek time. So does this mean that for HDTune in this case the STR and seek time data are kind of misrepresented and should be taken separately rather than read as being on the same part of the drive?

Well - if I understand your question correctly - yes. As it does not show which area of the platter is the 0% and which is the 100%... and even if - like HDTach - you won't be able to change the way the drive runs (from the outside inward) so in that case 0GB = outer edge, max GB = inner edge.