- Status
silverblue
Distinguished
- Jul 22, 2009
- 1,199
- 4
- 19,285
[citation][nom]djdarko321[/nom]im not gonna repeat myself soz. but i do recommend you read ALL my postings on this link here @ Tom's ofc then post back here logically and correct. I DO WISH INTEL WAS CHEAPER FO SHO! but, due to market shananigans that Intel loves to play, we wont see AMD as aggressively "original" back when they took that crappy proc and gave it an IMC removing the FSB. nvm im not gonna type all that again soz. READ ALL of what i posted tope of 4th page on this link and all the way down XD tnx[/citation]
Post logically and correct...? You are the one who said x86-64 was an iteration of IA-64. Nakal correctly pointed out this foible and you then retorted that he couldn't read... and in any case, Wikipedia is mostly correct and well maintained AND provides citations, and the internet would be poorer for its loss. The threat to go and edit the article in question is a little peurile, you must admit.
You're accusing me of fanboyism yet you're the one slating one company in favour of the other. I want good processors from both companies, but I favour the underdog. Sure, they bought the technology which eventually became the Athlon, but it won them a hell of a lot of market share as well as fans. Does that make me a fanboy? No. I considered a Core 2 Duo setup before setting on a PII X3 setup; the price, plus that extra core, won it for me. That's my right as a consumer - to choose for myself. It's a shame that AMD doesn't advertise because they may get more money which in turn would help drive their own innovation. Please remember that AMD do innovate, they don't just imitate. A period of stagnation doesn't mean they're sitting still. When their next models appear, we'll get some better competition and lower prices - we win, regardless of whoever has the best processors.
[citation][nom]silverblue[/nom]If each AMD core had an HT-style implementation, a 12-thread AMD CPU, assuming it was properly fed.[/citation]
I forgot to finish this part earlier; if this implementation existed, it'd make for a much more capable processor. However, it'd probably still fall well short of Gulftown due to Nehalem's superior workrate and extra cache. AMD didn't increase Thuban's cache and I believe it suffers for it, and there's always the question of a triple channel IMC.
Intel charges what it likes because AMD don't have a processor range of equal standing. If they did, Intel would be forced to slash the pricing of most of their models, especially the i7-980X. Sure, people do say that you get what you pay for, but let's consider that their flagship costs more than three times of AMD's for an additional 30% or so performance on average, plus AMD offers more for the money at nearly every price you can think of. Intel are selling vast quantities using the sheer weight of their name.
AMD is continuing with the Phenom II for the time being, and I'm sure it's hurting them to do so especially at 45nm, but if they can continue to refine their processes and stay competitive on the pricing front, the money will keep rolling in. It's important that they don't rush Bulldozer. It is also a shame that Llano doesn't sport Bulldozer cores. You'd expect even a dual-core Bulldozer CPU to outpace a quad-core STARS at the same clock speed, but we won't find that out for about a year.
I'm extremely interested in how powerful Sandy Bridge is, though I'm a little sceptical that it'll result in a performance increase similar to that of Nehalem over Penryn. We shall see!
Post logically and correct...? You are the one who said x86-64 was an iteration of IA-64. Nakal correctly pointed out this foible and you then retorted that he couldn't read... and in any case, Wikipedia is mostly correct and well maintained AND provides citations, and the internet would be poorer for its loss. The threat to go and edit the article in question is a little peurile, you must admit.
You're accusing me of fanboyism yet you're the one slating one company in favour of the other. I want good processors from both companies, but I favour the underdog. Sure, they bought the technology which eventually became the Athlon, but it won them a hell of a lot of market share as well as fans. Does that make me a fanboy? No. I considered a Core 2 Duo setup before setting on a PII X3 setup; the price, plus that extra core, won it for me. That's my right as a consumer - to choose for myself. It's a shame that AMD doesn't advertise because they may get more money which in turn would help drive their own innovation. Please remember that AMD do innovate, they don't just imitate. A period of stagnation doesn't mean they're sitting still. When their next models appear, we'll get some better competition and lower prices - we win, regardless of whoever has the best processors.
[citation][nom]silverblue[/nom]If each AMD core had an HT-style implementation, a 12-thread AMD CPU, assuming it was properly fed.[/citation]
I forgot to finish this part earlier; if this implementation existed, it'd make for a much more capable processor. However, it'd probably still fall well short of Gulftown due to Nehalem's superior workrate and extra cache. AMD didn't increase Thuban's cache and I believe it suffers for it, and there's always the question of a triple channel IMC.
Intel charges what it likes because AMD don't have a processor range of equal standing. If they did, Intel would be forced to slash the pricing of most of their models, especially the i7-980X. Sure, people do say that you get what you pay for, but let's consider that their flagship costs more than three times of AMD's for an additional 30% or so performance on average, plus AMD offers more for the money at nearly every price you can think of. Intel are selling vast quantities using the sheer weight of their name.
AMD is continuing with the Phenom II for the time being, and I'm sure it's hurting them to do so especially at 45nm, but if they can continue to refine their processes and stay competitive on the pricing front, the money will keep rolling in. It's important that they don't rush Bulldozer. It is also a shame that Llano doesn't sport Bulldozer cores. You'd expect even a dual-core Bulldozer CPU to outpace a quad-core STARS at the same clock speed, but we won't find that out for about a year.
I'm extremely interested in how powerful Sandy Bridge is, though I'm a little sceptical that it'll result in a performance increase similar to that of Nehalem over Penryn. We shall see!
masterofevil22
Distinguished
- May 13, 2010
- 229
- 0
- 18,690
Kudos to a well thought out and non-biased perspective.
Competition is the fuel that drives innovation and we are lucky as consumers to have it.
Competition is the fuel that drives innovation and we are lucky as consumers to have it.
djdarko321
Distinguished
- May 12, 2010
- 40
- 0
- 18,530
people always take things out of context, no matter how its worded. and yes Silverblue, if ur a professional u know that Wikipedia is not ur friend. That wasnt a threat to change, but as a statement of truth, ANYONE can post whatever on it. Then again only people that never went to college would use it. Kind of a shame tho that AMD just NOW caught up to the Penryn C2Q's. And slating one company over the other? LMAO no! I loved the Athlon's when they came out. It was the IMC that saved AMD's ass back then tho. We need more originality, or another company who can be. yes Nehalem uses a triple channel IMC (QPI), Sandy Bridge will use a four channel IMC (QPI). Hopefully on an all new design, not another revamp like Intel did by basically adding triple channel IMC (QPI)bringing back HT and adding some optimiations to the instruction sets of the Core 2's. And AMD sitting still? lolz What happened when they introduced their TRUE QUAD core (the first Phenoms)? FLOP....but thats what happens when the markets put this "budget" mindset on the people. They think yes , a less performing system w more cores tho for cheaper. LMAO
I said it before .... do what i do. run DAW's w VSTi's and raw audio and video design. watch em choke....... that is all tnx.
I said it before .... do what i do. run DAW's w VSTi's and raw audio and video design. watch em choke....... that is all tnx.
silverblue
Distinguished
- Jul 22, 2009
- 1,199
- 4
- 19,285
Phenom was an evolution of the K8 architecture and not a brand new architecture as is the case with Bulldozer. Compared to Core 2 Duo, Phenom isn't anywhere near as different from the previous models.
AMD is talking about the successor to Bulldozer but we're talking 2015 before it appears. That's how long it can take to bring out the next big thing, especially if you lack the resources of your main competitor. Bulldozer has been in development for a long time. AMD's method of allowing you to use a specific processor on different chipsets has probably saved them, along with their die harvesting, however had Intel not been so involved in anti-competitive practices, AMD may have had Bulldozer out earlier... and had much more money to invest in R&D and promotion.
As for Wikipedia... I once read an encyclopaedia and it said that the largest shark in the world was the Great White. It's extremely fashionable to knock Wikipedia, hell I once saw that somebody had edited an article on David James (England goalkeeper) to show he was 4 1/2 metres tall, and another person had put the Whale Shark down as "12.65km long and a weight of 21.5 million ton pounds". The editing history was clear to see and was quickly corrected in both cases. Wikipedia's not perfect, but there are people out there who care enough to correct its many mistakes (a lot of which are made by the malicious). In any case...
[citation][nom]djdarko321[/nom]Then again only people that never went to college would use it.[/citation]
...can't really be taken out of context. I'd like to think that I was educated at some point in my life.
Back to the Athlon, AMD didn't just add an IMC to make the AMD64, but they did get lucky because Intel got complacent and spent the next few years playing catch-up. Clock-for-clock, AMD was easily ahead in most things. Nowadays, it's the other way around. It'll be much harder for AMD to reverse that trend like Intel did, however they will at least give it a go.
Phenom's main issues were the TLB bug, CnQ not working at all well with Vista and its 65nm process. The first two should've been dealt with far earlier, the latter was obviously out of their control. I'm not sure the lack of L3 cache was a massive issue because the Athlon II X3s and X4s aren't far behind their Phenom II cousins at the same clock speed. In the end though, had Phenom not just been a revision of the K8, it would almost certainly have been a stronger performer.
AMD is talking about the successor to Bulldozer but we're talking 2015 before it appears. That's how long it can take to bring out the next big thing, especially if you lack the resources of your main competitor. Bulldozer has been in development for a long time. AMD's method of allowing you to use a specific processor on different chipsets has probably saved them, along with their die harvesting, however had Intel not been so involved in anti-competitive practices, AMD may have had Bulldozer out earlier... and had much more money to invest in R&D and promotion.
As for Wikipedia... I once read an encyclopaedia and it said that the largest shark in the world was the Great White. It's extremely fashionable to knock Wikipedia, hell I once saw that somebody had edited an article on David James (England goalkeeper) to show he was 4 1/2 metres tall, and another person had put the Whale Shark down as "12.65km long and a weight of 21.5 million ton pounds". The editing history was clear to see and was quickly corrected in both cases. Wikipedia's not perfect, but there are people out there who care enough to correct its many mistakes (a lot of which are made by the malicious). In any case...
[citation][nom]djdarko321[/nom]Then again only people that never went to college would use it.[/citation]
...can't really be taken out of context. I'd like to think that I was educated at some point in my life.
Back to the Athlon, AMD didn't just add an IMC to make the AMD64, but they did get lucky because Intel got complacent and spent the next few years playing catch-up. Clock-for-clock, AMD was easily ahead in most things. Nowadays, it's the other way around. It'll be much harder for AMD to reverse that trend like Intel did, however they will at least give it a go.
Phenom's main issues were the TLB bug, CnQ not working at all well with Vista and its 65nm process. The first two should've been dealt with far earlier, the latter was obviously out of their control. I'm not sure the lack of L3 cache was a massive issue because the Athlon II X3s and X4s aren't far behind their Phenom II cousins at the same clock speed. In the end though, had Phenom not just been a revision of the K8, it would almost certainly have been a stronger performer.
silverblue
Distinguished
- Jul 22, 2009
- 1,199
- 4
- 19,285
djdarko321
Distinguished
- May 12, 2010
- 40
- 0
- 18,530
agreed. 
but you have to admit tho Silverblue, if we had more competitors, other than the frivolous monopolistic Intel and the poor old AMD who came to life mainly w help from IBM in the court BS to clone the 386 (AM386), we wouldnt be hindered technologically with these market fiascoes that put budget in our minds when in reality its older tech by a few years. u think i enjoyed paying for my production system? but , gotta love Intel for that one .... (sarcastically speaking) wish there were cap laws for monopoly industries like Intel or something to that aspect. It'd force them to be on the ball and cheaper whilst allowing the other companies to to more directly compete with them across the board. Yes I love the fact u can keep same AMD board for a few years and just keep throwing new cpu's in them, but what forced them to do that? INTEL.... (their market BS sux and i hate em for it, but their 980X proc performs great for what i need and do) too bad NVidia wont jump on the cpu bandwagon cuz thatd rattle Intel fo sho, or IBM make a Cell (PS3) CPU platform for the PC markets. lolz Intel holds most of the cards which sux due to instruction set licensing. Fab this and blah blah like they are w the Global Foundries BS theyre trying to pull. basically due to the market BS we are forced to reap these benefits of the "budget" mindset and settle for tech from 3-5 years ago. Sux that a system I need costed almost 4G's USD $$. AMD needs help from somewheres thats for sure to help em push out their roadmap faster. Wish IBM bought em LMAO
but you have to admit tho Silverblue, if we had more competitors, other than the frivolous monopolistic Intel and the poor old AMD who came to life mainly w help from IBM in the court BS to clone the 386 (AM386), we wouldnt be hindered technologically with these market fiascoes that put budget in our minds when in reality its older tech by a few years. u think i enjoyed paying for my production system? but , gotta love Intel for that one .... (sarcastically speaking) wish there were cap laws for monopoly industries like Intel or something to that aspect. It'd force them to be on the ball and cheaper whilst allowing the other companies to to more directly compete with them across the board. Yes I love the fact u can keep same AMD board for a few years and just keep throwing new cpu's in them, but what forced them to do that? INTEL.... (their market BS sux and i hate em for it, but their 980X proc performs great for what i need and do) too bad NVidia wont jump on the cpu bandwagon cuz thatd rattle Intel fo sho, or IBM make a Cell (PS3) CPU platform for the PC markets. lolz Intel holds most of the cards which sux due to instruction set licensing. Fab this and blah blah like they are w the Global Foundries BS theyre trying to pull. basically due to the market BS we are forced to reap these benefits of the "budget" mindset and settle for tech from 3-5 years ago. Sux that a system I need costed almost 4G's USD $$. AMD needs help from somewheres thats for sure to help em push out their roadmap faster. Wish IBM bought em LMAOdjdarko321
Distinguished
- May 12, 2010
- 40
- 0
- 18,530
the IMC got AMD the performance crown. no more middle man as i stated elsewheres. (FSB - front side bus) Core 2 arch still used a FSB and smoked ALL Athlons. Now AMD has passed the C2Q's (Finally) They NEED help.
silverblue
Distinguished
- Jul 22, 2009
- 1,199
- 4
- 19,285
The first step will be Llano.
You're using a system that cost nearly $4K, but how much would it have cost for the highest performing AMD setup, and how much slower would it be? Having said that, if I had the money, I'd surely have gone the same way for now.
AM2 was released long before Core 2 Duo came out, and whereas AMD may not have meant to stretch its life out so long with AM2+ and AM3, the fact is it was a very good decision.
Phenom is relatively old, yes, however as you said yourself, Nehalem is also derivative. By definition, it's not much newer than Phenom, plus it uses an IMC as well as an L3 cache. If software was better threaded and/or more friendly towards AMD processors, the gap wouldn't be so flattering to Intel. Like you said though, it's not new tech. Phenom II was the result of some fixes plus a smaller manufacturing process that AMD simply had to use in order to remain competitive (and alive, it has to be said). AMD should've done more about the mobile market when it became evident that the 45nm process paid real dividends and dumped the older K8 tech in favour for Phenom II far quicker than it did - we all saw how good the CULV Core 2s were. They also should've gone with low-K a lot earlier, if possible - the PII X4 965 doesn't exactly sip power.
You're using a system that cost nearly $4K, but how much would it have cost for the highest performing AMD setup, and how much slower would it be? Having said that, if I had the money, I'd surely have gone the same way for now.
AM2 was released long before Core 2 Duo came out, and whereas AMD may not have meant to stretch its life out so long with AM2+ and AM3, the fact is it was a very good decision.
Phenom is relatively old, yes, however as you said yourself, Nehalem is also derivative. By definition, it's not much newer than Phenom, plus it uses an IMC as well as an L3 cache. If software was better threaded and/or more friendly towards AMD processors, the gap wouldn't be so flattering to Intel. Like you said though, it's not new tech. Phenom II was the result of some fixes plus a smaller manufacturing process that AMD simply had to use in order to remain competitive (and alive, it has to be said). AMD should've done more about the mobile market when it became evident that the 45nm process paid real dividends and dumped the older K8 tech in favour for Phenom II far quicker than it did - we all saw how good the CULV Core 2s were. They also should've gone with low-K a lot earlier, if possible - the PII X4 965 doesn't exactly sip power.
djdarko321
Distinguished
- May 12, 2010
- 40
- 0
- 18,530
when dealing with audio/video editing and production, the difference is like night and day. My friend got the 1055T 6 core and has it OC'd to 3.4ghz using 8 gb ddr3, along with the same maudio audiophile 2496 card as me. I took 2 of my four 295's and put em in his setup. (since his psu wouldnt power all 4 of mine)device manager seen all 4 GPU's. back to what im sayin, he does audio production like i do (Cubase 5 with various VSTi's and other samplers and midi devices) it took twice as long for him to open a project of mine on his setup let alone when playing it w all channels (Cubase audio channels midi and raw audio)playing, the cpu gauge in Cubase 5 was pegged and started stuttering like mad no matter how we adjusted the latencies on the sound card. I hooked my system back up and let him play with it LMAO he didnt want to leave. Thats just in audio, not video with Nuendo, Adobe Master Bundle CS4, or Maya. Big difference in instruction set optimizations between the two families of cpu's! so yes VERY HUGE noticeable differences between them.
IMO Adobe needs to get with the times. I wonder if they are partially owned by Intel at times considering they are soo optimized towards Intel cpus. http://www.hardwarecanucks.com/forum/hardware-canucks-reviews/31407-amd-phenom-ii-x6-1055t-1090t-six-core-processors-review-8.html
Look at the chart, even the dual core Intel 661 outperforms Intel's quads. Adobe benefits only from IPC wich is why it doesn't bode well with an AMD cpu. Adobe is not multithreaded friendly, even the 980x is barely faster than the 870 quad.
Biggest issue with programs like this is it makes the cpu look worthless when infact its the programming thats worthess. 64 bit has been out for how long now and Adobe doesn't even have a compatible flash player?
Look at the chart, even the dual core Intel 661 outperforms Intel's quads. Adobe benefits only from IPC wich is why it doesn't bode well with an AMD cpu. Adobe is not multithreaded friendly, even the 980x is barely faster than the 870 quad.
Biggest issue with programs like this is it makes the cpu look worthless when infact its the programming thats worthess. 64 bit has been out for how long now and Adobe doesn't even have a compatible flash player?
djdarko321
Distinguished
- May 12, 2010
- 40
- 0
- 18,530
Adobe is the bottom of the barrel for optimizations, security of using their products, and not fully utilizing the current instruction set/mutlticore technologies out atm YES. But the professional high end software like the Steinberg products I use or Maya, and the other various things i do, utilizes these techs. So the gains are very much noticeable. Just depends on the user and what u rly need tbh. but with the current marketing BS goin on, that has been for some time now, (INTEL) we will suffer for quite awhile to come.
tharkis842
Distinguished
- Apr 2, 2009
- 95
- 5
- 18,635
Kind of odd placement in the cpu race. Hopefully this will help bring some prices down *crosses fingers*.
noobinberg
Distinguished
- Jan 27, 2009
- 44
- 0
- 18,530
[citation][nom]mosu[/nom]Wonder why 875K is not a 32nm upgrade, we could see some serious overclocking at low power consumption.Still waiting for affordable (for me)pure processors without graphics with 4/8 cores on 32nm.[/citation]
Ok...I'm building DAW's
I'd like to see something like what the performance is using Sonar 8 OR ProTools LE with 24 audio tracks and pushing 4 effect plugins (preferably WAVE) on each track. Just play it and if you don't get an audio drop out just keep adding plugin effects to the tracks till it does. My current box chokes at 8 tracks and 4 effects per track on playback. But I can at least record the audio for those 8 tracks at once. Then do a bounce to disk test...how long to bounce the 24 tracks to 7.1 24bit WAV for movie soundtrack.
Music production is very demanding on the cpu. To record a full orchestra you could tax pretty much any cpu to it's limit. That's why the studios are buying new dual Xeon quad cores, (However macs haven't made the sweet jump to Nahelem architecture yet...which makes me question their inflated pricing, but that's a whole nuther can worms.)
I'm interested in making my next DAW rig based on an Intel or AMD Windows environment. To have the comparison done with software that I already use would definitely give me an idea of where to go.
Ok...I'm building DAW's
I'd like to see something like what the performance is using Sonar 8 OR ProTools LE with 24 audio tracks and pushing 4 effect plugins (preferably WAVE) on each track. Just play it and if you don't get an audio drop out just keep adding plugin effects to the tracks till it does. My current box chokes at 8 tracks and 4 effects per track on playback. But I can at least record the audio for those 8 tracks at once. Then do a bounce to disk test...how long to bounce the 24 tracks to 7.1 24bit WAV for movie soundtrack.
Music production is very demanding on the cpu. To record a full orchestra you could tax pretty much any cpu to it's limit. That's why the studios are buying new dual Xeon quad cores, (However macs haven't made the sweet jump to Nahelem architecture yet...which makes me question their inflated pricing, but that's a whole nuther can worms.)
I'm interested in making my next DAW rig based on an Intel or AMD Windows environment. To have the comparison done with software that I already use would definitely give me an idea of where to go.
noobinberg
Distinguished
- Jan 27, 2009
- 44
- 0
- 18,530
Oh...the quotes go with the post below them....oops!
The quote I meant to use said something about giving some test suggestions...
The quote I meant to use said something about giving some test suggestions...
djdarko321
Distinguished
- May 12, 2010
- 40
- 0
- 18,530
[citation][nom]noobinberg[/nom]Ok...I'm building DAW'sI'd like to see something like what the performance is using Sonar 8 OR ProTools LE with 24 audio tracks and pushing 4 effect plugins (preferably WAVE) on each track. Just play it and if you don't get an audio drop out just keep adding plugin effects to the tracks till it does. My current box chokes at 8 tracks and 4 effects per track on playback. But I can at least record the audio for those 8 tracks at once. Then do a bounce to disk test...how long to bounce the 24 tracks to 7.1 24bit WAV for movie soundtrack.Music production is very demanding on the cpu. To record a full orchestra you could tax pretty much any cpu to it's limit. That's why the studios are buying new dual Xeon quad cores, (However macs haven't made the sweet jump to Nahelem architecture yet...which makes me question their inflated pricing, but that's a whole nuther can worms.)I'm interested in making my next DAW rig based on an Intel or AMD Windows environment. To have the comparison done with software that I already use would definitely give me an idea of where to go.[/citation]
Although Sonar is mediocre at best for production (compared to Cubase5, Logic or Pro Tools HD), check these articles from Cakewalk themselves
http://blog.cakewalk.com/cakewalk-takes-the-stage-at-the-intel-developer-forum/
thats on an i7 965 at that
http://www.cakewalk.com/PCResource/Articles/ProcSpotlight.aspx
Cakewalk's recommended BEST lol if more is needed holla! i dont mind sharing info and links to such facts
Although Sonar is mediocre at best for production (compared to Cubase5, Logic or Pro Tools HD), check these articles from Cakewalk themselves
http://blog.cakewalk.com/cakewalk-takes-the-stage-at-the-intel-developer-forum/
thats on an i7 965 at that
http://www.cakewalk.com/PCResource/Articles/ProcSpotlight.aspx
Cakewalk's recommended BEST lol if more is needed holla! i dont mind sharing info and links to such facts
djdarko321
Distinguished
- May 12, 2010
- 40
- 0
- 18,530
u find me a professional audio production/visual design studio that has something other than "Intel Inside". Theres a reason for it
silverblue
Distinguished
- Jul 22, 2009
- 1,199
- 4
- 19,285
Okay, okay, we get the message. I should point out, however, that Cakewalk is optimised specifically for the i7. As is pretty much everything else.
What is ever optimised specifically for AMD? I suppose, in the end, that's AMD's issue. AMD still don't have a model with SSE4.1 and that must be hurting them.
In the end though, you buy what you need in order to get the job done. Not everyone needs a system for the same tasks and not everyone has their CPU(s) at 100% constantly. AMD needs to replace Deneb with Zosma ASAP and maximise the money they're getting in whilst they're preparing Bulldozer - if they can't win on performance, at least attempt to win on platform cost and power usage as compared to the 45nm Nehalems and Lynnfields (not exactly hard, really).
What is ever optimised specifically for AMD? I suppose, in the end, that's AMD's issue. AMD still don't have a model with SSE4.1 and that must be hurting them.
In the end though, you buy what you need in order to get the job done. Not everyone needs a system for the same tasks and not everyone has their CPU(s) at 100% constantly. AMD needs to replace Deneb with Zosma ASAP and maximise the money they're getting in whilst they're preparing Bulldozer - if they can't win on performance, at least attempt to win on platform cost and power usage as compared to the 45nm Nehalems and Lynnfields (not exactly hard, really).
djdarko321
Distinguished
- May 12, 2010
- 40
- 0
- 18,530
ye true the only sse4 implementation AMD has and uses is only 4 instuctions total, (SSE4a)
LZCNT Leading Zero Count - bit manipulation.
LZCNT instruction may also be implemented in some processors that do not support the other SSE4 instructions and a separate bit can be tested to confirm LZCNT presence.
POPCNT Population count (count number of bits set to 1). POPCNT instruction may also be implemented in some processors that do not support the other SSE4 instructions and a separate bit can be tested to confirm POPCNT presence.
EXTRQ/INSERTQ Combined mask-shift instructions.
MOVNTSD/MOVNTSS Scalar streaming store instructions.
SSE4.1 SSE4.2 intel uses MPSADBW Compute eight offset sums of absolute differences (i.e. |x0-y0|+|x1-y1|+|x2-y2|+|x3-y3|, |x0-y1|+|x1-y2|+|x2-y3|+|x3-y4|, ...); this operation is extremely important for modern HDTV codecs, and (see [5]) allows an 8x8 block difference to be computed in fewer than seven cycles. One bit of a three-bit immediate operand indicates whether y0 .. y10 or y4 .. y14 should be used from the destination operand, the other two whether x0..x3, x4..x7, x8..x11 or x12..x15 should be used from the source.
PHMINPOSUW Sets the bottom unsigned 16-bit word of the destination to the smallest unsigned 16-bit word in the source, and the next-from-bottom to the index of that word in the source.
PMULDQ Packed signed multiplication on two sets of 2 out of 4 packed integers, the 1st and 3rd per packed 4, giving 2 packed 64-bit results.
PMULLD Packed signed multiplication, 4 packed sets of 32-bit integers multiplied to give 4 packed 32-bit results.
DPPS, DPPD Dot product for AOS (Array of Structs) data. This takes an immediate operand consisting of four (or two for DPPD) bits to select which of the entries in the input to multiply and accumulate, and another four (or two for DPPD) to select whether to put 0 or the dot-product in the appropriate field of the output.
BLENDPS, BLENDPD, BLENDVPS, BLENDVPD, PBLENDVB, PBLENDW Conditional copying of elements in one location with another, based (for non-V form) on the bits in an immediate operand, and (for V form) on the bits in register XMM0.
PMINSB, PMAXSB, PMINUW, PMAXUW, PMINUD, PMAXUD, PMINSD, PMAXSD Packed minimum/maximum for different integer operand types
ROUNDPS, ROUNDSS, ROUNDPD, ROUNDSD Round values in a floating-point register to integers, using one of four rounding modes specified by an immediate operand
INSERTPS, PINSRB, PINSRD/PINSRQ, EXTRACTPS, PEXTRB, PEXTRW, PEXTRD/PEXTRQ The INSERTPS and PINSR instructions read 8, 16 or 32 bits from an x86 register memory location and insert it into a field in the destination register given by an immediate operand, EXTRACTPS and PEXTR read a field from the source register and insert it into an x86 register or memory location. For example, PEXTRD eax, [xmm0], 1; EXTRACTPS [addr+4*eax], xmm1, 1 stores the first field of xmm1 in the address given by the first field of xmm0.
PMOVSXBW, PMOVZXBW, PMOVSXBD, PMOVZXBD, PMOVSXBQ, PMOVZXBQ, PMOVSXWD, PMOVZXWD, PMOVSXWQ, PMOVZXWQ, PMOVSXDQ, PMOVZXDQ Packed sign/zero extension to wider types
PTEST This does the same as the TEST instruction, in that it sets the ZF and CF flags to the result of an AND between its operators ... it sets the Z flag if any of the bits matched, and the C flag if all of them did.
PCMPEQQ Quadword (64 bits) compare for equality
PACKUSDW Convert signed DWORDs into unsigned WORDs with saturation.
MOVNTDQA Efficient read from write-combining memory area into SSE register; this is useful for retrieving results from peripherals attached to the memory bus.
SSE4.2
These instructions were first implemented in the Intel Core i7 product line and complete the SSE4 instruction set.
Instruction Description
CRC32 Accumulate CRC32C value using the polynomial 0x11EDC6F41 (or, without the high order bit, 0x1EDC6F41).[6]
PCMPESTRI Packed Compare Explicit Length Strings, Return Index
PCMPESTRM Packed Compare Explicit Length Strings, Return Mask
PCMPISTRI Packed Compare Implicit Length Strings, Return Index
PCMPISTRM Packed Compare Implicit Length Strings, Return Mask
PCMPGTQ Compare Packed Signed 64-bit data For Greater Than
POPCNT Population count (count number of bits set to 1). POPCNT instruction may also be implemented in some processors that do not support the other SSE4 instructions and a separate bit can be tested to confirm POPCNT presence.
AMD made sse5 which intel made their iteration of it called AVX .... instruction sets do everything....makes the cpu handle the types of data differently as they get processed
LZCNT Leading Zero Count - bit manipulation.
LZCNT instruction may also be implemented in some processors that do not support the other SSE4 instructions and a separate bit can be tested to confirm LZCNT presence.
POPCNT Population count (count number of bits set to 1). POPCNT instruction may also be implemented in some processors that do not support the other SSE4 instructions and a separate bit can be tested to confirm POPCNT presence.
EXTRQ/INSERTQ Combined mask-shift instructions.
MOVNTSD/MOVNTSS Scalar streaming store instructions.
SSE4.1 SSE4.2 intel uses MPSADBW Compute eight offset sums of absolute differences (i.e. |x0-y0|+|x1-y1|+|x2-y2|+|x3-y3|, |x0-y1|+|x1-y2|+|x2-y3|+|x3-y4|, ...); this operation is extremely important for modern HDTV codecs, and (see [5]) allows an 8x8 block difference to be computed in fewer than seven cycles. One bit of a three-bit immediate operand indicates whether y0 .. y10 or y4 .. y14 should be used from the destination operand, the other two whether x0..x3, x4..x7, x8..x11 or x12..x15 should be used from the source.
PHMINPOSUW Sets the bottom unsigned 16-bit word of the destination to the smallest unsigned 16-bit word in the source, and the next-from-bottom to the index of that word in the source.
PMULDQ Packed signed multiplication on two sets of 2 out of 4 packed integers, the 1st and 3rd per packed 4, giving 2 packed 64-bit results.
PMULLD Packed signed multiplication, 4 packed sets of 32-bit integers multiplied to give 4 packed 32-bit results.
DPPS, DPPD Dot product for AOS (Array of Structs) data. This takes an immediate operand consisting of four (or two for DPPD) bits to select which of the entries in the input to multiply and accumulate, and another four (or two for DPPD) to select whether to put 0 or the dot-product in the appropriate field of the output.
BLENDPS, BLENDPD, BLENDVPS, BLENDVPD, PBLENDVB, PBLENDW Conditional copying of elements in one location with another, based (for non-V form) on the bits in an immediate operand, and (for V form) on the bits in register XMM0.
PMINSB, PMAXSB, PMINUW, PMAXUW, PMINUD, PMAXUD, PMINSD, PMAXSD Packed minimum/maximum for different integer operand types
ROUNDPS, ROUNDSS, ROUNDPD, ROUNDSD Round values in a floating-point register to integers, using one of four rounding modes specified by an immediate operand
INSERTPS, PINSRB, PINSRD/PINSRQ, EXTRACTPS, PEXTRB, PEXTRW, PEXTRD/PEXTRQ The INSERTPS and PINSR instructions read 8, 16 or 32 bits from an x86 register memory location and insert it into a field in the destination register given by an immediate operand, EXTRACTPS and PEXTR read a field from the source register and insert it into an x86 register or memory location. For example, PEXTRD eax, [xmm0], 1; EXTRACTPS [addr+4*eax], xmm1, 1 stores the first field of xmm1 in the address given by the first field of xmm0.
PMOVSXBW, PMOVZXBW, PMOVSXBD, PMOVZXBD, PMOVSXBQ, PMOVZXBQ, PMOVSXWD, PMOVZXWD, PMOVSXWQ, PMOVZXWQ, PMOVSXDQ, PMOVZXDQ Packed sign/zero extension to wider types
PTEST This does the same as the TEST instruction, in that it sets the ZF and CF flags to the result of an AND between its operators ... it sets the Z flag if any of the bits matched, and the C flag if all of them did.
PCMPEQQ Quadword (64 bits) compare for equality
PACKUSDW Convert signed DWORDs into unsigned WORDs with saturation.
MOVNTDQA Efficient read from write-combining memory area into SSE register; this is useful for retrieving results from peripherals attached to the memory bus.
SSE4.2
These instructions were first implemented in the Intel Core i7 product line and complete the SSE4 instruction set.
Instruction Description
CRC32 Accumulate CRC32C value using the polynomial 0x11EDC6F41 (or, without the high order bit, 0x1EDC6F41).[6]
PCMPESTRI Packed Compare Explicit Length Strings, Return Index
PCMPESTRM Packed Compare Explicit Length Strings, Return Mask
PCMPISTRI Packed Compare Implicit Length Strings, Return Index
PCMPISTRM Packed Compare Implicit Length Strings, Return Mask
PCMPGTQ Compare Packed Signed 64-bit data For Greater Than
POPCNT Population count (count number of bits set to 1). POPCNT instruction may also be implemented in some processors that do not support the other SSE4 instructions and a separate bit can be tested to confirm POPCNT presence.
AMD made sse5 which intel made their iteration of it called AVX .... instruction sets do everything....makes the cpu handle the types of data differently as they get processed
silverblue
Distinguished
- Jul 22, 2009
- 1,199
- 4
- 19,285
All of the above is copied directly from http://en.wikipedia.org/wiki/SSE4 - pardon me but you can see the irony.
Other than that, we're in agreement about instruction sets. Bulldozer should address that deficiency.
Other than that, we're in agreement about instruction sets. Bulldozer should address that deficiency.
djdarko321
Distinguished
- May 12, 2010
- 40
- 0
- 18,530
so if cakewalk, Steinberg, pro tools and logic is optimised to i7's how come they've ALWAYS ran better w Intel inside? mac even? lol
my answer...
Intel's instruction sets allow them to. It seems that Intel wouldnt license SSE4.1 and SSE4.2 to amd, so amd made SSE4a which is just a fraction of ALL the codings.
my answer...
Intel's instruction sets allow them to. It seems that Intel wouldnt license SSE4.1 and SSE4.2 to amd, so amd made SSE4a which is just a fraction of ALL the codings.
- Status
TRENDING THREADS
-
-
-
Question Download speed is slow on only my PC - please help!- Started by Ograsmium
- Replies: 16
-
-
Question New installed W11 on sdd, D drive (HDD) goes to the 100% in seconds
- Started by Emily_Vin
- Replies: 7
Facebook
Twitter
Instagram