View unanswered posts | View active topics It is currently Thu Mar 28, 2024 12:58 pm



Reply to topic  [ 10 posts ] 
 ARM vs X86 – Key differences explained! 
Author Message

Joined: Wed Jan 16, 2013 2:33 am
Posts: 165
ARM vs X86 – Key differences explained!

Quote:
When it comes to 64-bit computing, there are also some significant differences between ARM and Intel. Did you know that Intel didn’t even invent the 64-bit version of its x86 instruction set.


http://www.androidauthority.com/arm-vs- ... ed-568718/


Tue Dec 29, 2015 3:31 pm
Profile

Joined: Sat Aug 22, 2015 6:26 am
Posts: 40
Interesting article, but I'm not sure I agree with the statement
Quote:
ARM’s big.LITTLE architecture is an innovation that Intel is nowhere near replicating.

I don't see any particular reason why Intel couldn't do something similar.


Tue Dec 29, 2015 8:17 pm
Profile

Joined: Wed Jan 16, 2013 2:33 am
Posts: 165
I'm sure they are both capable companies with different markets so differences exist.
Maybe the writer has a reason or maybe it could be bias; I don't know.

Intel and AMD have had a long history in court:

http://www.cnet.com/news/intel-and-amd- ... -in-court/

This other article is enjoyable but it shows the underhandedness in the market and it could cause bias if you are on one side or the other:

http://hackaday.com/2015/12/09/echo-of- ... then-lost/

What do we know?

ARM has been doing good in the mobile and low power markets and they have a lot going on for them.


Tue Dec 29, 2015 9:02 pm
Profile

Joined: Wed Jan 09, 2013 6:54 pm
Posts: 1780
Interesting: ARM and Intel have very similar net income per employee. Very different businesses. ARM has somewhat higher growth. Their licensees ship over ten billion ARMs a year. Intel sells maybe a twentieth in quantity but of course at much higher prices.

I have a soft spot for ARM because of their 6502 beginnings, and because they are a UK success story.

On a much more technical front, for some remarkable optimisations going on right now in a 6502 emulator written for the ARM-powered Raspberry Pi, see the last few pages of this thread:
http://stardot.org.uk/forums/viewtopic. ... &start=240
Yesterday the emulator was running at 150MHz, now it's at 163MHz or so. On a 700MHz ARM. These last few optimisations have come from deep study of the pipeline behaviour.


Tue Dec 29, 2015 9:27 pm
Profile

Joined: Sat Aug 22, 2015 6:26 am
Posts: 40
Quote:
ARM has been doing good in the mobile and low power markets and they have a lot going on for them.

Indeed, but as they're trying to increase the performance, the power consumption will have to go up too. At the same time, Intel is approaching the same target from the other side. I don't see any bottleneck in the two architectures. They are both capable.

Biggest selling point for ARM is the licensing deal, where third parties can make their own SoC. Biggest selling point for Intel is their closely controlled manufacturing process that's still a step ahead of everybody else's.

Best thing for the consumer would be if they keep competing.


Tue Dec 29, 2015 11:46 pm
Profile

Joined: Tue Jan 15, 2013 10:11 am
Posts: 114
Location: Norway/Japan
BigEd wrote:
On a much more technical front, for some remarkable optimisations going on right now in a 6502 emulator written for the ARM-powered Raspberry Pi, see the last few pages of this thread:
http://stardot.org.uk/forums/viewtopic. ... &start=240
Yesterday the emulator was running at 150MHz, now it's at 163MHz or so. On a 700MHz ARM. These last few optimisations have come from deep study of the pipeline behaviour.
Yes, I've been following that thread during the Xmas holidays, and it's nearly unbelievable that you can have a 163MHz 6502 on that board. I didn't think it would have gone any further after it reached 150MHz (and I think I saw bets to that effect mentioned), but then again I thought it was impressive to start with, at around 70MHz IIRC?
Maybe there actually is still, deep down, a connection to the 6502 architecture from the ARM.. as e.g. the flags and some other bits could be handled nearly natively.


Wed Dec 30, 2015 3:38 pm
Profile

Joined: Wed Jan 09, 2013 6:54 pm
Posts: 1780
Indeed so, the flags are very similar - it's not too surprising, given Acorn's previous history with the 6502. It's not even impossible, given Acorn's business plan to make a 32-bit RISC computer to replace the BBC Micro, that they had some ideas about emulating the 6502 even while designing the ARM.

As we're in an ARM-related thread, it might be worth mentioning a couple of blogs which are analysing the ARM1 implementation as seen in the visualARM simulation:
http://daveshacks.blogspot.co.uk/
http://www.righto.com/search/label/reverse-engineering

(And, for completeness, a sketch of ARM's licensing model:
http://www.anandtech.com/show/7112/the- ... el-works/2
)


Thu Dec 31, 2015 8:30 pm
Profile

Joined: Tue Jan 15, 2013 10:11 am
Posts: 114
Location: Norway/Japan
And now it's at 176MHz.. incredible.


Fri Jan 01, 2016 10:32 am
Profile

Joined: Tue Feb 10, 2015 7:07 am
Posts: 52
Tor wrote:
And now it's at 176MHz.. incredible.

Actually, those MHz figures are slightly misleading.

There are several versions of the BBC Basic benchmark, calibrated for different versions of BBC Basic.

The one that I have been using is calibrated for Basic II, but the BBC Master I'm testing on is running Basic IV. In BBC Basic IV the floating point code is much more efficient, hence the tendency for the Benchmark to over-state the equivalent 65C02 frequency in MHz.

This morning I dug around a bit and found the version that's correctly calibrated for Basic IV:
Attachment:
IMG_0205.JPG
IMG_0205.JPG [ 385.36 KiB | Viewed 9320 times ]

If we can get to 140MHz on this benchmark, then a 700MHz ARM will be emulating a 6502 at 1/5 of the clock speed.

Dave


Fri Jan 01, 2016 2:58 pm
Profile

Joined: Wed Jan 09, 2013 6:54 pm
Posts: 1780
(Amusing side-note on x86 vs ARM: the x86 instruction set architecture runs to 2900 pages. ARM is of course a much simpler machine with less legacy, so it's only 2700 pages. via RISC-V presentation, via lowRISC conf which has an interesting pointer to "Pydgin a fast and productive instruction-set simulator".)


Fri Jan 08, 2016 3:52 pm
Profile
Display posts from previous:  Sort by  
Reply to topic   [ 10 posts ] 

Who is online

Users browsing this forum: No registered users and 8 guests


You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot post attachments in this forum

Search for:
Jump to:  
cron
Powered by phpBB® Forum Software © phpBB Group
Designed by ST Software