Author |
Message |
|
Quertior
Newbie
Joined: 30 Aug 2009 Posts: 12
|
Posted: 31 Aug 2009 09:17:34 pm Post subject: |
|
|
Hello all,
I have been wondering lately, what is the fastest currently available TI calc?
I mean fastest as in "Calculation speed" i.e. given a certain CPU-intensive program, which calculator could complete it the fastest?
I am mainly interested in calcs other than the Nspire, because
1) We don't even know for sure what proc is in the Nspire;
2) It currently can only exec BASIC programs;
3) It is currently TI's only calc armed with an ARM processor.
So, given those rules, which calc would be the fastest?
Also, please feel free to correct me if I'm wrong about the nspire.
Last edited by Guest on 31 Aug 2009 09:45:14 pm; edited 1 time in total |
|
Back to top |
|
|
JoeYoung
Advanced Member
Joined: 15 Nov 2008 Posts: 316
|
Posted: 31 Aug 2009 09:41:57 pm Post subject: |
|
|
89t?
just guessing, though. |
|
Back to top |
|
|
FloppusMaximus
Advanced Member
Joined: 22 Aug 2008 Posts: 472
|
Posted: 31 Aug 2009 09:49:34 pm Post subject: |
|
|
In terms of pure CPU power... well, it's hard to compare different processor architectures, but the 68ks will presumably be faster for most tasks. I think the HW4 Titanium has the fastest CPU of the bunch.
But there's a lot more to it than that. If you're interested in graphics performance, for instance, note that you can get higher frame rates for full-screen animation if your screen is smaller. If you're interested in advanced math functions, or BASIC performance, you need to take the OS into account. (As a rule, older OSes seem to be better optimized and faster than newer ones. And the Z80 OSes, despite their many flaws, seem to be considerably more optimized than the 68k ones. Funny thing, that.)
And most people programming for the Z80 use hand-written assembly language, whereas most people programming for the 68k use a C compiler, so we Z80 folks may have a slight advantage there.
Last edited by Guest on 31 Aug 2009 09:55:27 pm; edited 1 time in total |
|
Back to top |
|
|
Quertior
Newbie
Joined: 30 Aug 2009 Posts: 12
|
Posted: 31 Aug 2009 09:57:53 pm Post subject: |
|
|
Mmmmmkay thanks for the replies. I currently have 3 calcs: 84+, 84PSE, and an HW4 89T. I have an 86 in the mail, should arrive soon
So what about the z80 calcs? What is the fastest among them? I know the 84PSE has the fastest clock speed, so that would presumably mean it is the fastest? Unless the OS is at fault, I suppose. |
|
Back to top |
|
|
FloppusMaximus
Advanced Member
Joined: 22 Aug 2008 Posts: 472
|
Posted: 31 Aug 2009 10:10:23 pm Post subject: |
|
|
I doubt you'll find much difference between the 84+ and 84+ SE. The 83+ (SE) OS is a little faster since it contains less cruft, but it doesn't seem to make much difference in practice. I actually don't know how the CPU speeds compare. |
|
Back to top |
|
|
Quertior
Newbie
Joined: 30 Aug 2009 Posts: 12
|
Posted: 31 Aug 2009 10:19:36 pm Post subject: |
|
|
FloppusMaximus wrote: I actually don't know how the CPU speeds compare.
I looked it up...
The procs in the original 83 and 83+ are 6MHz.
The proc in the 83PSE is 15MHz.
The 84+ and SE both have 15MHz procs, with a 6MHz compatibility mode. (The 83PSE may have this mode too) |
|
Back to top |
|
|
FloppusMaximus
Advanced Member
Joined: 22 Aug 2008 Posts: 472
|
Posted: 31 Aug 2009 10:38:59 pm Post subject: |
|
|
Well, yes, I know that the speeds are the same on paper, but none of the calculators run at exactly 6 MHz or 15 MHz. |
|
Back to top |
|
|
Quertior
Newbie
Joined: 30 Aug 2009 Posts: 12
|
Posted: 31 Aug 2009 10:43:12 pm Post subject: |
|
|
FloppusMaximus wrote: Well, yes, I know that the speeds are the same on paper, but none of the calculators run at exactly 6 MHz or 15 MHz.
What are you implying? I assume Z80 fabs operate the same as fabs for regular computer chips, in that, as the line matures, it produces more stable and generally "better" parts. Is this correct? |
|
Back to top |
|
|
Michael
Newbie
Joined: 21 Dec 2007 Posts: 39
|
Posted: 01 Sep 2009 12:19:49 am Post subject: |
|
|
FloppusMaximus wrote: And most people programming for the Z80 use hand-written assembly language
Unless you're TI and you are releasing an app called CellSheet.
I also enjoy pointing out that the 83+ SE was originally going to run at 25 MHz (modes for 6, 15, 20, and 25) but never made it that way to production.
Last edited by Guest on 01 Sep 2009 12:22:21 am; edited 1 time in total |
|
Back to top |
|
|
DrDnar
Member
Joined: 28 Aug 2009 Posts: 116
|
Posted: 01 Sep 2009 02:38:17 pm Post subject: |
|
|
Michael wrote: I also enjoy pointing out that the 83+ SE was originally going to run at 25 MHz (modes for 6, 15, 20, and 25) but never made it that way to production.
I've always wondered, whence did you get that information?
Last edited by Guest on 01 Sep 2009 02:41:56 pm; edited 1 time in total |
|
Back to top |
|
|
fullmetalcoder
Member
Joined: 01 Aug 2009 Posts: 139
|
Posted: 01 Sep 2009 02:53:04 pm Post subject: |
|
|
There are three values of "fast mode" : 01, 02 and 03.
To each of these values correspond a different LCD delay port which indicates that they were meant to differ by more than a mere hundredth of MHz (estimation found on WikiTI).
Last edited by Guest on 01 Sep 2009 02:53:42 pm; edited 1 time in total |
|
Back to top |
|
|
DrDnar
Member
Joined: 28 Aug 2009 Posts: 116
|
Posted: 01 Sep 2009 04:23:15 pm Post subject: |
|
|
Yes, I know that. I wrote a program to compute CPU speed, which you can find on TI-Calc at http://www.ticalc.org/archives/files/fileinfo/420/42000.html .
Michael recently released bunch of his notes, source code, and other stuff. One of the documents appears to be documentation on the crystal timers. It mentions "when FCLK[1:0]=11 (~25MHz)", "when FCLK[1:0]=10 (~20MHz)", and "when FCLK[1:0]=11 (~15MHz)". It's possible this is where he got the information, but it's also possible he has other documents he isn't giving us. (I'm assuming the document is from TI because it uses such technical language.) But all this only makes me wonder where he got those documents. |
|
Back to top |
|
|
fullmetalcoder
Member
Joined: 01 Aug 2009 Posts: 139
|
Posted: 01 Sep 2009 04:49:55 pm Post subject: |
|
|
[quote name='Dr. D'nar' post='136179' date='Sep 1 2009, 11:23 PM']Michael recently released bunch of his notes, source code, and other stuff.[/quote]
where? |
|
Back to top |
|
|
DrDnar
Member
Joined: 28 Aug 2009 Posts: 116
|
Posted: 01 Sep 2009 05:40:52 pm Post subject: |
|
|
http://www.radicalsoft.org/michael/misc.zip He posted the link in the comments in his retirement TI Calc article.
I say screw the NDAs. What else are they hiding?
On topic:
The 83+SE OS is a little faster, as Flop said, since it has less stuff in it. The the Z80 calcs effectively contain the same Z80 CPU, so the only difference in performance is the CPU clock and software. If you'd like to know how the TI-83+SE and the TI-83+/SE compare in raw CPU speed, get my program and run it on different models with the same batteries. (The CPU clocks likely vary with battery level.) Of course, you can always overclock your TI-83+ and MAKE it the fastest . |
|
Back to top |
|
|
Michael
Newbie
Joined: 21 Dec 2007 Posts: 39
|
Posted: 01 Sep 2009 09:24:15 pm Post subject: |
|
|
I actually obtained that timer document only after the crystal timers had already been reverse engineered, making it amusingly rather useless to me (but perhaps part of the reason it was obtainable). That's mostly what I base my 25 MHz statement on, along with other circumstantial evidence. I don't have any other significant documents, nor did I ever sign any NDAs.
Last edited by Guest on 01 Sep 2009 09:30:45 pm; edited 1 time in total |
|
Back to top |
|
|
FloppusMaximus
Advanced Member
Joined: 22 Aug 2008 Posts: 472
|
Posted: 01 Sep 2009 09:57:43 pm Post subject: |
|
|
Wow, that timer documentation is fascinating. I knew a lot of the stuff in there already, but I had no idea about the connection with port 2F. |
|
Back to top |
|
|
|