< Back to OSY 1.0 thread list

OSY 1.0 Thread Viewer

Thread #: 1644

Sucktacular

AllYorBaseRBelong2Us

Wed Mar 27 22:50:59 2002

The performance advantage of the PowerPC G4 starts with its data pipeline. The term “processor pipeline” refers to the number of processing steps, or stages, it takes to accomplish a task. The fewer the steps, the shorter — and more efficient — the pipeline. Thanks to its efficient 7-stage design (versus 20 stages for the Pentium 4 processor) the G4 processor can accomplish a task with 13 fewer steps than the PC. You do the math.

[url]http://www.apple.com/powermac/processor.html[/url]

That is indeed some nice propaganda they've got going on. :)

i like how when technical facts stare Apple threateningly in the face that they are able to put their sophestry to work to make it all look in their favor.  Faaaaaaaantasstesticular.

Harbinger

Thu Mar 28 01:36:22 2002

Well, that just completely justifies 18 months at 500MHz, doesn't it? ;)
AllYorBaseRBelong2Us

Thu Mar 28 01:58:55 2002

Well, the good part about it is that it fuels the fires if iTardome that is always so much fun to combat.  Just ask Hitscan. :cheesy:
Harbinger

Thu Mar 28 02:23:26 2002

LOL.  Xspung's bag of apples is a good plan. ;)
OscarWilde

Thu Mar 28 03:16:59 2002

<AOL> You got mail </AOL>
Imitation Gruel

Thu Mar 28 03:39:19 2002

The more propaganda I read from Apple, the more I am convinced that anyone who believes it must have approximately the brain functionality and maturity of a flatworm.
OscarWilde

Thu Mar 28 03:48:55 2002

The more propaganda I read from Apple, Microsoft, and the linux community, the more I am convinced that anyone who believes it must have approximately the brain functionality and maturity of a flatworm.
AllYorBaseRBelong2Us

Thu Mar 28 03:56:50 2002

teeheehee!

That was some funny smam OW. I shall return the favor tonight, after some AEO II

(Edited by AllYorBaseRBelong2Us at 9:57 pm on Mar. 27, 2002)

HitScan

Thu Mar 28 03:59:36 2002

Just ask Hitscan. :cheesy:

Well that's twice you've mentioned my foray into the real life battlefield. Yes, let's do ask HitScan. :biggrin:

I like Apple
:beret:     :)
         That's great. I don't think they're too bad either.

You don't have enough Apples here. You suck cowballs.
:beret:     :(
         What the hell brought that on you 'tard?

Later...

Ahh! the pain!
:bash:
     Take it! Take it like a little bitch!

:cheesy:

:beret: <- Best Smiley EVAR. :cheesy:

(Edited by HitScan at 8:01 pm on Mar. 27, 2002)

OscarWilde

Thu Mar 28 04:00:57 2002

LOL@hitscan.com
AllYorBaseRBelong2Us

Thu Mar 28 04:02:01 2002

Bravo Hitscan! :)

That was a nice little smiley skit.

i hope you don't mind me mentioning you, I was just trying to give you a hard time about it :D

All in good fun!

HitScan

Thu Mar 28 04:03:52 2002

LOL@hitscan.com

Hm? A slip of the keyboard, or a funneh intarweb euphamism? Because I sadly do not own hitscan.com, if you were in fact referring to that for some reason. :)

AYB: It's quite alright. Had you not, I'd have not done this lovely episode of "Smiley Theatre" :biggrin:

AllYorBaseRBelong2Us

Thu Mar 28 04:04:58 2002

incidently, I just tried hitscan.com and nothing came up.

HitScan

Thu Mar 28 04:09:57 2002

I'm thinking of grabbing .net sometime, but until I get some permanent IPs it does me little good, heh.

I must do Smiley Theatre more often. That can be some funny shit. :) Sadly, it's a PITA to line up. (especially with no <tt> type tags. Eww. :biggrin: )

OscarWilde

Thu Mar 28 04:10:43 2002

i'd put the money on teh funneh intraweb euphamism

;)

DrPizza

Thu Mar 28 05:12:15 2002

You do the math.

OK.

P4 2.2 GHz:
SPECint2000: 784
SPECfp2000: 777

G4 1GHz:
TOO FUCKING EMBARASSED TO SUBMIT ANY SCORES

'nuff said.

AllYorBaseRBelong2Us

Thu Mar 28 05:34:31 2002

from DrPizza posted at 11:12 pm on Mar. 27, 2002

You do the math.

OK.

P4 2.2 GHz:
SPECint2000: 784
SPECfp2000: 777

G4 1GHz:
TOO FUCKING EMBARASSED TO SUBMIT ANY SCORES

'nuff said.

For once, Apple's one step ahead of you Pizza.

Two paragraphs later:

Another aspect of speculative operation worth noting is that it is possible to create (for testing purposes) a contrived set of instructions that can make the processor guess correctly much more often than it would under real-world conditions. Thus a “benchmark” with no relation to actual performance can be crafted to cleverly avoid the bubble problem and thus indicate unrealistically high performance. This underscores the importance of using real applications to provide valid performance comparisons.

An obvious, but not blatant stab at SPEC, however innacurate the picture they are trying to paint of what SPEC is.

Which is of course especially shamefull given the BYTEmark they swore by three-four years ago.

It's amazing.  It reminds me of a kid that punches his brother in the face, and then runs to mother (when she comes to investagate the sound of crying) accusing his brother of acually of doing the deed to *him*

Wow!  That's bold and funny :cheesy:

DrPizza

Thu Mar 28 05:45:38 2002

An obvious, but not blatant stab at SPEC, however innacurate the picture they are trying to paint of what SPEC is.

Which is of course especially shamefull given the BYTEmark they swore by three-four years ago.

It's amazing.  It reminds me of a kid that punches his brother in the face, and then runs to mother (when she comes to investagate the sound of crying) accusing his brother of acually of doing the deed to *him*

Wow!  That's bold and funny


LOL.

So, SPEC's no good, but determining speed from pipeline depth's A-OK?

:rolleyes:

It's pitiful.

DrPizza

Thu Mar 28 05:46:57 2002

And whatever SPEC's failings, using unrealistic algorithms isn't really one of them.

Stupid Apple cunts.

AllYorBaseRBelong2Us

Thu Mar 28 05:53:37 2002

Speaking of which:

[url=http://www.newtek.com/discus/messages/2/16932.html?1013655265]133 MHz bus means nothing, except in comparison to other PowerPC chips with different bus speed. The throughput (bandwidth) of the bus is what matters.

The Athlon XP has a 200 MHz double clocked bus and uses DDR DRAM -- but can only move 700 MB/s (on a good day).

The P4 has a fast bus and dual channel RDRAM and can move 1500 MB/s -- but only on very simple operations like memcpy, memset, memcmp. For complex operations it's not much better than PC133 (where it only moves about 600 MB/s).

The PowerMac G4 has a lowly 133 MHz bus, and moves 1085 MB/s, sustained over large (32 Meg) buffers. For less optimized code it can sustain 930 MB/s (again over large buffers).

Why is the slowest bus moving so much memory? Better bus design, better DRAM controller, better cache design, and several other details that only serious solderheads would understand. [/url]

Sounds like LB's good buddy is bending things a bit.  What he's quoting for the Power Mac IIRC, is at the limit of PC133's bandwidth.

I may start a BF discussion on this subject to see what others have to say about this to try to find out what ccox is talking about.

(Edited by AllYorBaseRBelong2Us at 11:55 pm on Mar. 27, 2002)

Xspung

Thu Mar 28 05:55:25 2002

So, SPEC's no good, but determining speed from pipeline depth's A-OK?

Huh??? Apple's talking about SPEC in that quote?

I thought they were speaking of Steve's MacWorld Photoshop bakeoff.

OscarWilde

Thu Mar 28 06:59:37 2002

[url=http://www.mackido.com/Myths/performance_techie.html]You are wrong! POWERMAC OWNZ j00!!!! I know so because I read it here!!!!![/url]


MUAHAHAHAHAHAHAhA!!!!!!!!!!!!!

I have brought the wrath of David K. Every upon this forum!!!

muahahahahahhahahaha!!!!!

/madan gets 0.04 cents richer!

just incase you silly head don't get it, me joke about mackido

AllYorBaseRBelong2Us

Thu Mar 28 07:46:35 2002

Hehehe.


Silly poopy-headed tech evangelist :cheesy:

HitScan

Thu Mar 28 14:39:12 2002

Eww. a DKE article linked on OSY. How will we get the smell out now? :(
AllYorBaseRBelong2Us

Thu Mar 28 17:47:12 2002

Eww. a DKE article linked on OSY. How will we get the smell out now?

Egads no.  Everyone Knows DKE is a shill that wouldn't know what he's talking about in such things (remember, Protected Memory is Bad!).  The above quote is from Jeremy's best pal, Cris Cox. :)

Harbinger

Thu Mar 28 18:06:30 2002

from HitScan posted at 9:39 am on Mar. 28, 2002

Eww. a DKE article linked on OSY. How will we get the smell out now? :(

I'm thinking that lots of harsh chemicals are needed.

DrPizza

Fri Mar 29 05:01:28 2002

The figures he gives for the Mac's memory throughput are hilarious.

They're above the theoretical maximum throughput for PC133 memory -- a figure the G4 couldn't hope to reach anyway, if what Billy Goat Gruff is true, because one in five cycles is a control cycle anyway.

Tee hee.

Evil Merlin

Fri Mar 29 11:21:28 2002

But it HAS to be true! It was on the net!
HitScan

Fri Mar 29 13:16:23 2002

teh intarweb dos not lye!
Imitation Gruel

Fri Mar 29 14:32:47 2002

Hmm, those statements make me :eek:.

That guy DKE is a real :asshole: for saying such stupid shit. I'm like, :wtf: here.

Admittedly, I'm no :angel:, but Macolytes like that guy are the type I just want to :bash: like this.

DKE: :madfawk: :madfawk: :madfawk:.

(Edited by Imitation Gruel at 6:33 am on Mar. 29, 2002)

AllYorBaseRBelong2Us

Fri Mar 29 17:31:02 2002

Oh, there's the DKE article.  Poopyhead indeed ulr'ed a DKE story.

Sucktacular :cheesy:

Imitation Gruel

Fri Mar 29 18:56:48 2002

Linking to DKE's site should be punishable by a bad "temporary" custom title.

Like Ars member clintatpurdue's title: I made Ohrmazd mad, now my bum is stinky

AllYorBaseRBelong2Us

Thu Apr 4 21:05:40 2002

Here's another item:

Looking through the various Motorola PDF's,  I have come to the following conclusion:

how the fark can the PPC7450 sustain anywhere near it's advertised max gigaflops throughput when it's Mem->L3->L2->L1 bandwidth is only 64-bits wide?  

In order to sustain this, Altivec could only be running it's operands out of it's registers, or only one operand from L1 at a time (L1-> Registers is a 128bit wide path)

This I would like to know!

AllYorBaseRBelong2Us

Fri Apr 5 05:25:08 2002

YOu see, I was always under the impression that the L2->L1 interface was 256bits wide.

I appear to have been wrong.  Forget yer DDR ram, if this isn't fixed, i predict that ddr systems will not have much benifit for the PowerPC.

OscarWilde

Fri Apr 5 09:32:56 2002

from AllYorBaseRBelong2Us posted at 4:05 am on April 5, 2002

Here's another item:

Looking through the various Motorola PDF's,  I have come to the following conclusion:

how the fark can the PPC7450 sustain anywhere near it's advertised max gigaflops throughput when it's Mem->L3->L2->L1 bandwidth is only 64-bits wide?  

In order to sustain this, Altivec could only be running it's operands out of it's registers, or only one operand from L1 at a time (L1-> Registers is a 128bit wide path)

This I would like to know!

Well assuming that the Mem->L3->l2->l1 is all running at the same speed but:
Mem SDR:133mhz
L3 DDR: So i guess 266mhz?
L2 at processor speed
L1 at processor speed

OscarWilde

Fri Apr 5 09:37:00 2002

from AllYorBaseRBelong2Us posted at 12:25 pm on April 5, 2002

YOu see, I was always under the impression that the L2->L1 interface was 256bits wide.

I appear to have been wrong.  Forget yer DDR ram, if this isn't fixed, i predict that ddr systems will not have much benifit for the PowerPC.

Accoring to [url=http://e-www.motorola.com/webapp/sps/site/prod_summary.jsp?code=MPC7450&nodeId=03M943030450467M98653] motorola[/url] your first impression was right:


Cache and MMU Support

The MPC7450 processor has separate 32KB, physically addressed instruction and data caches. Both L1 caches feature cache way locking and are eight-way set associative. For greater speed, the L2 cache has been integrated on-chip with a 256-bit interface to L1 which operates at processor frequency.

ph34r my tagnut skillzzz!!!

:cheesy:

OscarWilde

Fri Apr 5 09:50:00 2002

AYB, i'm confused at how you and I are looking at Motorola's info but  I am reading different specs then you are?

I'm looking through the pdf files incase there is something more to the story but so far I can't find anything that supports what you saying?
:confused:

Anyway check [url=http://e-www.motorola.com/brdata/PDFDB/docs/MPC7450TS.pdf]this technical summary pdf file[/url].

Then goto page 4 of 52 and look at the block diagram. I can't makes no tails of it, but maybe you can understand it better.

OscarWilde

Fri Apr 5 09:55:47 2002

on page 9 the "L1 is fully pipelined to provide 128bits/cycle to and from the VR"
"The L2 is fully Pipelines to provide 256bit per processor cycle to the L1"

Now i'm curious to how the PPC chip is configured in my powermac.

AllYorBaseRBelong2Us

Fri Apr 5 10:04:58 2002

L3 DDR: So i guess 266mhz?

The L3 speed varies according to processor speed rather than bus speed.  IIRC.

ph34r my tagnut skillzzz!!!

I fear your poopiness more :)

AYB, i'm confused at how you and I are looking at Motorola's info but  I am reading different specs then you are?

I'm looking through the pdf files incase there is something more to the story but so far I can't find anything that supports what you saying?

I was looking at the MPC7450TS.pdf yesterday trying to visualize the layout when the L1-L2 question popped into my head.  I read the PDF on Cache latencies which somehow lead me to this conclusion.

it appear I was wrong though, but damn, I wish they'd make it more clear in the diagram :rolleyes:

BTW, the diagram is for the 7451, which must differ from the 7450 somehow.

ph43r my stinkiness! :cheesy:

AllYorBaseRBelong2Us

Fri Apr 5 10:09:09 2002

What would I do if you weren't around to keep me straight :)

I'd prolly poop myself shitless :cheesy:

AllYorBaseRBelong2Us

Sat Apr 6 00:26:27 2002

Well, the old G4 L2->L1 bandwidth was indeed 64-bit.

unless I am mistaken again.

Of course I am [url=http://e-www.motorola.com/brdata/PDFDB/docs/MPC7410EC.pdf]not[/url]

:)

OscarWilde

Sat Apr 6 06:34:18 2002

Dude stay consistent. You mentioned the 7450 (and the 7451 is not a major revision over the 7450 so the specs still apply). Then you talk about the 7400 differences as if to prove a point. I know about the 7400/10 variants. The interface might be 64bit but the speed between each different levels of memory are different.
I forget what the setup is, but the speed between the L2 and L1 for the 7400 is 2:1 or 1:1 which means half or same as processor speed. Some of them had higher fractions but that depends on which powermac model you are talking about.
AllYorBaseRBelong2Us

Sat Apr 6 07:20:53 2002

Then you talk about the 7400 differences as if to prove a point. I know about the 7400/10 variants. The interface might be 64bit but the speed between each different levels of memory are different.

I wasn't trying to prove anything, I was merely Yammering on like a ravenouse buffoon, it's not like I'm anti PPC.  :tongue:

I forget what the setup is, but the speed between the L2 and L1 for the 7400 is 2:1 or 1:1 which means half or same as processor speed. Some of them had higher fractions but that depends on which powermac model you are talking about

Yes indeed. :)

OscarWilde

Sat Apr 6 08:50:34 2002

Thats what I'm wondering about. Why did you bring up the 7450? At first i was assuming it had to do with the Adobe cox comments but I knew that was about the 7400 chip. Even then i didn't believe cox's comments held much water. I assume then you were curious about the current powermac line and if they could achieve the gigaflops advertised so i just jumped in like yer poopy headed pal to sort out the specs. Then you talk about the 7400. So for me  it was kinda like, HUH! what are we discussing here? Have you been sniffing the Shakira CD too much?
;) :cheesy:
AllYorBaseRBelong2Us

Sat Apr 6 09:58:56 2002

Thats what I'm wondering about. Why did you bring up the 7450?

Because it relates the the link I originally posted.  I think we are so spoiled with every thread going off topic by the fifth reply that an on topic disscussion is disorienting us :cheesy:

At first i was assuming it had to do with the Adobe cox comments but I knew that was about the 7400 chip.

Actually I was refering to the newer 7450's originally when I thought that they had a 64bit L1-L2 bus.  To clarify, I was mistaken.

I assume then you were curious about the current powermac line and if they could achieve the gigaflops advertised so i just jumped in like yer poopy headed pal to sort out the specs. Then you talk about the 7400.

That's because I'm a silly Stinky-head :tongue:

teeheehee :)

it was kinda like, HUH! what are we discussing here? Have you been sniffing the Shakira CD too much?

I was daydreaming about snuggling with her :)