nixd2001
Sep 14, 07:48 PM
Originally posted by onemoof
Someone asked the difference between RISC and CISC.
First thing, there isn't that distinction anymore. RISC originally meant that the processor had fixed width instructions (so it wouldn't have to waste time asking the software how big the next instruction will be). CISC mean that the processor had variable width instructions (meaning time would have to be taken to figure out how long the next instruction is before fetching it.) However, Intel has addressed this problem by making it possible for the processor to switch to a fixed-width mode for special processor intensive purposes. The PowerPC is stuck with fixed-width and has no ability to enjoy the flexibility of variable-width instructions for non-processor-intensive tasks. This means that CISC is now better than RISC. (Using the terms to loosely define Pentium as CISC and PowerPC as RISC.)
Originally it was Reduced versus Complex instruction set computer. Making simpler processors go faster is generally easier than making complex processors go faster as there is less internal state/logic to synchronise and keep track of. For any given fabrication technology, this still generally holds true. Intel managed to sidestep this principle by investing massive sums in their fab plants, effectively meaning that the fab processes being compared weren't the same.
The opposite end of the spectrum from RISC is arguably the VAX line. With this instruction set, massive complexities arose from the fact that a single instruction took so long and did so much. It was possible for timers, interrupts and "page faults" to occur midway during an instruction. This required saving a lot of internal state so that it could later be restored. There were examples of performing a given operation with a single instruction or a sequence of instructions that performed the same effect, but where the sequence achieved the join quicker because the internal implementation within the processor was able to get on with the job quicker because it was actually a simpler task being asked of it.
The idea of fixed sized instructions isn't directly coupled to the original notion of RISC, although it is only one step behind. One of the basic ideas with the original RISC processors was that an instruction should only take a single cycle to complete. So a 100MHz CPU might actually achieve 100M instructions per second. (This was often not achieved due to memory latencies, but this isn't the "fault" of the processor core). In this context, having a variable length instruction means that it is easy for the instruction decoding (especially if it requires more than one "word") to require for effort than any other aspect of executing an instruction.
There are situations where a variable width instruction might have advantages, but the argument goes that breaking the overall task down into equal sized instructions means that fetching (including caching, branch predicting, ec) and decoding these instructions becomes simpler, permitting optimisations and speed gains to be made elsewhere in the processor design.
Intel blur RISC and CISC into gray by effectively executing RISC instructions internally, even if they support the apparent decoding of CISC insructions. They only do this for legacy reasons.
Apple will never switch to IA32 (Pentium) because 32 bit processors are a dead-end and maybe have a couple years left. The reason is because they can only have a maximum of 4 GB of RAM [ (2^32)/(1 Billion) = 4.29 GB ]. This limit is very close to being reached in current desktop computers. Apple MAY at some point decide to jump to IA64 in my opinion, and I think they should. Obviously the Intel family of processors is unbeatable unless they have some sort of catastrophe happen to them. If Apple jumped on they'd be back on track. Unfortunately I don't believe IA64 is yet cheap enough for desktops.
I think this "unbeatable" assertion requires some qualification. It may be that Intel will achieve the best price/performance ratio within a suitable range of qualifications, but this is different from always achieving best p/p ratio whatever. Indeed, IA64 versus Power4 is going to be an interesting battle because Intel has bet on ILP (instruction level parallelism) whereas IBM has bet on data bandwidth. Ultimately (and today!), I think IBM's bet has more going for it. But that's if you want ultimate performance. The PC space is often characterised by people apparenntly wanting ultimate performance but actually always massively qualifiying it with severe price restrictions (such as less than 5 digits to the price).
Someone asked the difference between RISC and CISC.
First thing, there isn't that distinction anymore. RISC originally meant that the processor had fixed width instructions (so it wouldn't have to waste time asking the software how big the next instruction will be). CISC mean that the processor had variable width instructions (meaning time would have to be taken to figure out how long the next instruction is before fetching it.) However, Intel has addressed this problem by making it possible for the processor to switch to a fixed-width mode for special processor intensive purposes. The PowerPC is stuck with fixed-width and has no ability to enjoy the flexibility of variable-width instructions for non-processor-intensive tasks. This means that CISC is now better than RISC. (Using the terms to loosely define Pentium as CISC and PowerPC as RISC.)
Originally it was Reduced versus Complex instruction set computer. Making simpler processors go faster is generally easier than making complex processors go faster as there is less internal state/logic to synchronise and keep track of. For any given fabrication technology, this still generally holds true. Intel managed to sidestep this principle by investing massive sums in their fab plants, effectively meaning that the fab processes being compared weren't the same.
The opposite end of the spectrum from RISC is arguably the VAX line. With this instruction set, massive complexities arose from the fact that a single instruction took so long and did so much. It was possible for timers, interrupts and "page faults" to occur midway during an instruction. This required saving a lot of internal state so that it could later be restored. There were examples of performing a given operation with a single instruction or a sequence of instructions that performed the same effect, but where the sequence achieved the join quicker because the internal implementation within the processor was able to get on with the job quicker because it was actually a simpler task being asked of it.
The idea of fixed sized instructions isn't directly coupled to the original notion of RISC, although it is only one step behind. One of the basic ideas with the original RISC processors was that an instruction should only take a single cycle to complete. So a 100MHz CPU might actually achieve 100M instructions per second. (This was often not achieved due to memory latencies, but this isn't the "fault" of the processor core). In this context, having a variable length instruction means that it is easy for the instruction decoding (especially if it requires more than one "word") to require for effort than any other aspect of executing an instruction.
There are situations where a variable width instruction might have advantages, but the argument goes that breaking the overall task down into equal sized instructions means that fetching (including caching, branch predicting, ec) and decoding these instructions becomes simpler, permitting optimisations and speed gains to be made elsewhere in the processor design.
Intel blur RISC and CISC into gray by effectively executing RISC instructions internally, even if they support the apparent decoding of CISC insructions. They only do this for legacy reasons.
Apple will never switch to IA32 (Pentium) because 32 bit processors are a dead-end and maybe have a couple years left. The reason is because they can only have a maximum of 4 GB of RAM [ (2^32)/(1 Billion) = 4.29 GB ]. This limit is very close to being reached in current desktop computers. Apple MAY at some point decide to jump to IA64 in my opinion, and I think they should. Obviously the Intel family of processors is unbeatable unless they have some sort of catastrophe happen to them. If Apple jumped on they'd be back on track. Unfortunately I don't believe IA64 is yet cheap enough for desktops.
I think this "unbeatable" assertion requires some qualification. It may be that Intel will achieve the best price/performance ratio within a suitable range of qualifications, but this is different from always achieving best p/p ratio whatever. Indeed, IA64 versus Power4 is going to be an interesting battle because Intel has bet on ILP (instruction level parallelism) whereas IBM has bet on data bandwidth. Ultimately (and today!), I think IBM's bet has more going for it. But that's if you want ultimate performance. The PC space is often characterised by people apparenntly wanting ultimate performance but actually always massively qualifiying it with severe price restrictions (such as less than 5 digits to the price).
Vertigo50
May 2, 12:49 PM
It's amazing what someone with the proper tool can do to actually test these kind of issues.
I was waiting for someone to FINALLY grab a caliper and measure both devices. Game over.
I was waiting for someone to FINALLY grab a caliper and measure both devices. Game over.
Natural Science
Mar 23, 10:50 PM
Will the Lion Server allow me to keep one set of data accessible from my iMac & MBP so they are basically working with only one set of files? I don't want to migrate data from my iMac to my new MBP because between the iLife projects I don't want them living on separate machines... I simply want to close iMovie or excel for example and pick right up where I left up on the MBP once I'm upstairs! I set up file sharing and accessing the iPhoto library from the iMac takes forever to load, nevermind the loss of certain features like location tagging and I've yet to get iMovie to open the iMac library without having it crash. I know I sound like an total moron here, but the good news is when it comes to computers, I am, and I've accepted that.;)
bond2
Sep 19, 01:44 PM
Everyone has been complaining so much about the current Macs being so far behind the Pentiums and AMD processors in speed. In actuality the new 1.25ghz Dual G4 is the 3rd fastest consumer PC in the market. It almost matches the fastest AMD processor and is just slightly behind Intels PIV 2.8 ghz machine. Check it out for yourself:
http://www.cpuscorecard.com/
http://www.cpuscorecard.com/
more...
syrianos
Sep 19, 04:31 PM
I'm running RAID0 too !
ulls derrick rose dunking
more...
chicago bulls derrick rose
chicago bulls derrick rose
more...
Derrick Rose #1 of the Chicago
GRINCH-derrick rose-ulls
more...
Tags: Derrick Rose, Joakim
chicago bulls derrick rose
more...
Derrick Rose #1 of the Chicago
chicago bulls derrick rose
more...
Images). OKLAHOMA CITY
Derrick Rose at Foot Locker
more...
Derrick Rose #1 Chicago Bulls
chicago bulls derrick rose
chicago bulls derrick rose
dicklacara
Mar 24, 10:32 AM
I'm all for more agencies adopting Macs! Heck, one day we might be able to classify them as switchers :D
Nah... based on military policy we wouldn't even hear of them...
...you know... Switch But Don't Tell!
:eek:
Nah... based on military policy we wouldn't even hear of them...
...you know... Switch But Don't Tell!
:eek:
more...
toddybody
Apr 21, 01:51 PM
Well, Looks like I know who to beat and rob.
Watch out Carmack...;)
Watch out Carmack...;)
jimsowden
Mar 28, 08:52 AM
Appception
more...
strabes
Mar 23, 01:50 PM
how bout let's make sure that airplay video playback works first before you try to license it out. has not work from day 1 on my appletv
Works perfectly for me (using Air Video and the latest beta server) since 4.3 came out.
Works perfectly for me (using Air Video and the latest beta server) since 4.3 came out.
SPUY767
Oct 28, 09:04 AM
This was one of the most handy little apps that I lost when OS X hit the scene. Super intuitive, super effective.
I have Peak and Pro Tools and neither handle those simple, everyday audio needs like a SoundEdit 16 could.
I very much think there is an audience for Soundbooth and that it will be well received. Based on my experienced with SoundEdit 16, if Soundbooth were Universal, I'd more than likely purchase... immediately.
Amen to that. Nothing I have ever found is as good at simple splicing of uncompressed audio files than SE16. Amadeus comes close, but not cigar, it just doesn't feel like SE16.
I have Peak and Pro Tools and neither handle those simple, everyday audio needs like a SoundEdit 16 could.
I very much think there is an audience for Soundbooth and that it will be well received. Based on my experienced with SoundEdit 16, if Soundbooth were Universal, I'd more than likely purchase... immediately.
Amen to that. Nothing I have ever found is as good at simple splicing of uncompressed audio files than SE16. Amadeus comes close, but not cigar, it just doesn't feel like SE16.
more...
Apple OC
May 2, 08:39 PM
Unless the microgenetics machines are small, well, there are several stages in DNA transcription and reverse transcriptase, which would make several conditions unsuitable. Heat destroys it. It can be awfully slow. There are at least 6 stages using different equipment.
They are not testing his DNA for a court of law ... just verifying that he is the one who deserves to be the dead guy
They are not testing his DNA for a court of law ... just verifying that he is the one who deserves to be the dead guy
DeSnousa
May 14, 07:13 PM
i mean what is $800 Australian in American dollars?
the reason i bring up the next GPU client is b/c i've heard it will be better with ATI cards than the current client. now i'm not sure if the ATI cards will pass nvidia or not though (i doubt it, but i'm just not sure).
right now intel is better than amd for folding, but that doesn't mean you can't fold with it.
with intel, if you go with let's say an i3 or i5 processor, you can always upgrade later. but keep in mind that i3, i5, and i7 860 use a different socket than the i7 920/930 and up.
Ahh stuff it I will spend roughly 900-1100 US dollars, so if you had that kind of money what would you get, don't need a screen, nor Windows 7. I appreciate your thought and help.
the reason i bring up the next GPU client is b/c i've heard it will be better with ATI cards than the current client. now i'm not sure if the ATI cards will pass nvidia or not though (i doubt it, but i'm just not sure).
right now intel is better than amd for folding, but that doesn't mean you can't fold with it.
with intel, if you go with let's say an i3 or i5 processor, you can always upgrade later. but keep in mind that i3, i5, and i7 860 use a different socket than the i7 920/930 and up.
Ahh stuff it I will spend roughly 900-1100 US dollars, so if you had that kind of money what would you get, don't need a screen, nor Windows 7. I appreciate your thought and help.
more...
levitynyc
Apr 1, 10:36 AM
Ditto! Slingbox rocks! And it's even more fun when you have a friend or two in other states that allow you to use their Slingbox to watch events that are blacked out in your own area. :)
Mark
Yup. If I ever switch providers I'm putting a Slingbox in my dad's house so I can watch the NFL Sunday Ticket.
You can pay the Yesnetwork.com 100 bucks to watch the Yankees online at work, or I can watch for free on my Slingbox.
Mark
Yup. If I ever switch providers I'm putting a Slingbox in my dad's house so I can watch the NFL Sunday Ticket.
You can pay the Yesnetwork.com 100 bucks to watch the Yankees online at work, or I can watch for free on my Slingbox.
Blue Velvet
Sep 13, 09:17 AM
Well, surgeons make me feel comfortable - or at least, it's easy to open up to them.
I know -- just a few minutes with them and they'll have you in stitches.
I know -- just a few minutes with them and they'll have you in stitches.
more...
jettredmont
Nov 21, 06:20 PM
If you're in a warm room, for instance, you'll have much lower performance, since it requires the differential to work. Of course, maybe the information available isn't wholly accurate, but that's my understanding based on the description.
And therein lies the failure of this idea as a simplifying concept:
When do you need the fan on? When the processor heats up.
Do you want the fan blowing harder or softer when the room is warmer? Harder.
In other words, if I'm sitting out in the cool evening air, I hardly need the fan going at all as the coolness of the air is doing just fine pulling the heat from the CPU. If I'm sitting in 100-degree weather then that fan better be buzzing like a bee to get enough air past the heat sink to effect a suitable heat transfer.
This works in just the opposite: In the cold air, there's a huge differential, so the fan is going full bore, annoying me and all my peace-and-quiet-loving neighbors. In the warm air, it slows to a crawl as the amount of electricity generated approaches the lower limit of sustaining power for the fan. Then it stops. Then my laptop heats up rapidly and the processor dies.
So, you need two additional controls: a bleed for cases when this extra cooling is not necessary, and a backup fan for when it isn't sufficient.
So, we haven't been able to simplify the problem at all, and instead are gaining the (very slight) power savings from not having to run this fan off our battery power (directly) in a mid-temp room. Seems like the R&D and per-unit costs put into this circuitry could be more wisely spent eking a few more milliwatts from the existing circuitry ...
And therein lies the failure of this idea as a simplifying concept:
When do you need the fan on? When the processor heats up.
Do you want the fan blowing harder or softer when the room is warmer? Harder.
In other words, if I'm sitting out in the cool evening air, I hardly need the fan going at all as the coolness of the air is doing just fine pulling the heat from the CPU. If I'm sitting in 100-degree weather then that fan better be buzzing like a bee to get enough air past the heat sink to effect a suitable heat transfer.
This works in just the opposite: In the cold air, there's a huge differential, so the fan is going full bore, annoying me and all my peace-and-quiet-loving neighbors. In the warm air, it slows to a crawl as the amount of electricity generated approaches the lower limit of sustaining power for the fan. Then it stops. Then my laptop heats up rapidly and the processor dies.
So, you need two additional controls: a bleed for cases when this extra cooling is not necessary, and a backup fan for when it isn't sufficient.
So, we haven't been able to simplify the problem at all, and instead are gaining the (very slight) power savings from not having to run this fan off our battery power (directly) in a mid-temp room. Seems like the R&D and per-unit costs put into this circuitry could be more wisely spent eking a few more milliwatts from the existing circuitry ...
Hans Brix
Apr 1, 02:43 AM
To be more specific, Costco gas was $3.939 for regular yesterday.
more...
iJon
Sep 15, 08:56 PM
As I understand it the drugs they use are poisons and the job of the anestisiologist (?) is to give you just enough to nearly kill you and keep that dose constant throughout the op, so your more dead than asleep.
Probably best not to try and fight with the guy trying very hard to not-quite-kill-you!
I guess I never really fight it. I'm out in a heartbeat. I just like to see who long it takes me, which is never long.
jon
Probably best not to try and fight with the guy trying very hard to not-quite-kill-you!
I guess I never really fight it. I'm out in a heartbeat. I just like to see who long it takes me, which is never long.
jon
cdebnil
Jan 4, 02:38 PM
Purchased and will try it out tonight. I'll try to make a video review of it in action.
InsanelyApple
Mar 13, 01:05 PM
My Verizon iPhone isn't experiencing this.
DoFoT9
Mar 1, 01:58 PM
Just looking at the ''Server Admin' screenshots posted, i noticed that in the list of Services, both AFP and NFS are missing. Is File Sharing managed somewhere else in Lion?
youre right! uh oh. will have to investigate when i get to work.
youre right! uh oh. will have to investigate when i get to work.
SilentPanda
Apr 11, 03:34 PM
I'm not finding it terribly hard although I think I'm on 2-3 right now. I did kind of get bored (but I was very tired last night, slept for 12 hours) during 2-3 and just shut the machine off without saving.
White for people who don't want to know *anything* although I'm still very vague --- I found the "collection" bit in 2-2 to be horrid and really think they should have made your task use the Wii controller in some fashion instead. I also dislike the seemingly "once per world invisible thing unless you point your Wii controller at the screen" thing they have going on. Lastly there really should be some way to quickly switch between characters and Pixls. ---
Overall the game is fun and sometimes quite interesting due to the use of 2D and 3D spaces but I really think they could have polished it more.
White for people who don't want to know *anything* although I'm still very vague --- I found the "collection" bit in 2-2 to be horrid and really think they should have made your task use the Wii controller in some fashion instead. I also dislike the seemingly "once per world invisible thing unless you point your Wii controller at the screen" thing they have going on. Lastly there really should be some way to quickly switch between characters and Pixls. ---
Overall the game is fun and sometimes quite interesting due to the use of 2D and 3D spaces but I really think they could have polished it more.
ArchaicRevival
Apr 21, 02:04 PM
Regardless, some people are still gonna complain of some unimportant missing feature.
drlunanerd
Oct 27, 08:16 AM
Lyra if you had read the thread you would have seen my comment about the student discount in the store.
Did they give you the full HE discount though - Leopard for �58.75? Or did they charge �75?
Did they give you the full HE discount though - Leopard for �58.75? Or did they charge �75?
gauchogolfer
Sep 25, 09:52 AM
www.tuaw.com
Reports release of Aperture 1.5
Details to follow.
Reports release of Aperture 1.5
Details to follow.
No comments:
Post a Comment