A computer components & hardware forum. HardwareBanter

If this is your first visit, be sure to check out the FAQ by clicking the link above. You may have to register before you can post: click the register link above to proceed. To start viewing messages, select the forum that you want to visit from the selection below.

Go Back   Home » HardwareBanter forum » Processors » Intel
Site Map Home Register Authors List Search Today's Posts Mark Forums Read Web Partners

Number of floating-point registers?



 
 
Thread Tools Display Modes
  #1  
Old September 8th 05, 07:39 PM
external usenet poster
 
Posts: n/a
Default Number of floating-point registers?

How many floating-point registers are available for use by programs in
any of Intel Celeron D Processors 340, 335, 330, 325, and 320?

  #2  
Old September 8th 05, 09:04 PM
Artie Ziff
external usenet poster
 
Posts: n/a
Default

On Thu, 08 Sep 2005 11:39:01 -0700, BigMan wrote:

How many floating-point registers are available for use by programs in
any of Intel Celeron D Processors 340, 335, 330, 325, and 320?


Only 8, architecturally. Not an actual limit as the FP (and integer)
registers are renamed to a much wider register file (say 32 FP regs
but the amount is implementation-specific).

  #4  
Old September 9th 05, 06:46 AM
external usenet poster
 
Posts: n/a
Default

Well, in fact I'm trying to understand the following:
1. What is the maximum number of 8-byte FP values that I can load into
registers before I start computations on these values.
In one case the computations are performed on 9 values, and in the
other - on 18.
2. Is it possible to perform computations on values already loaded into
registers while other values are being loaded into other registers?

  #5  
Old September 9th 05, 06:48 AM
external usenet poster
 
Posts: n/a
Default

One more thing - I guess the Celerons in question are not able to
operate in 64-bit mode.

  #6  
Old September 9th 05, 01:12 PM
Yousuf Khan
external usenet poster
 
Posts: n/a
Default

wrote:
Well, in fact I'm trying to understand the following:
1. What is the maximum number of 8-byte FP values that I can load into
registers before I start computations on these values.
In one case the computations are performed on 9 values, and in the
other - on 18.


Again, completely complicated by the fact that there are two floating
point mechanisms, x87 and SSE. x87 is stack-based (like an HP
calculator), while SSE is randomly accessible. Each type also has its
own seperate register set, and I don't think there is a way to copy or
share between the two register sets directly: you'll have to save the
value to memory from one set, and retrieve it from memory onto the
other. I may be wrong about the transfer mechanism I haven't kept
uptodate on it, but that's what I remember x87 and SSE registers can't
directly transfer to each other, except through the memory.

In either case, there are only eight registers whether it is x87 or SSE.
So the answer is no, you can't load all 9 or 18 values into registers to
begin computations on them, the job must be split up.

2. Is it possible to perform computations on values already loaded into
registers while other values are being loaded into other registers?


You'll have to get specific now. Which mechanism are you talking about
x87 or SSE? The answer is different in each case.

In x87, you typically load all of the values into the registers first,
and then begin computations. The computations are always done on the top
of the stack in a LIFO order. In SSE, you just do computations on
registers in any way you like, and you can load any empty registers
whenever you like, even if computations are going on other registers.

Yousuf Khan
  #8  
Old September 9th 05, 03:03 PM
external usenet poster
 
Posts: n/a
Default

I see that the information that I'm seeking is too much to be posted in
a newsgroup, so could you please, recommend some reading (whitepapers,
articles, or even books) about FP arithmetic on Intel, AMD and Apple
processors?

  #9  
Old September 9th 05, 04:34 PM
Yousuf Khan
external usenet poster
 
Posts: n/a
Default

http://developer.amd.com/documentation.aspx#devguides

 




Thread Tools
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

vB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Forum Jump

Similar Threads
Thread Thread Starter Forum Replies Last Post
Xbox360, PS3 : announcements of the death of PC gaming seem premature........ mace Nvidia Videocards 10 June 30th 05 08:25 PM
On the brink of madness... I.C. Koets General 18 January 31st 05 11:49 PM
Passmark Performance Test, Division, Floating Point Division, 2DShapes @(none) General 0 August 19th 04 11:57 PM
AMD64 vs. a floating point operation (FLOP) Only NoSpammers AMD x86-64 Processors 8 June 27th 04 03:55 PM
Drive Image 2002 Rosie Storage (alternative) 9 November 20th 03 04:25 PM


All times are GMT +1. The time now is 01:58 PM.


Powered by vBulletin® Version 3.6.4
Copyright ©2000 - 2024, Jelsoft Enterprises Ltd.
Copyright ©2004-2024 HardwareBanter.
The comments are property of their posters.