Slashdot Mirror
Dual-Core Shoot Out - Intel vs. AMD
sebFlyte writes "The Intel vs AMD battle of the benchmarks continues. ZDNet is running its rather comprehensive-looking guide to a side-by-side test of Intel and AMD's dual-core desktop chips, the Athlon 64 X2 3200+ and the Pentium D 820. They look at pure performance, as well as the difference it makes to apps you might use on the desktop. In the end, AMD comes out as the winner. From the article: 'AMD currently offers the most attractive dual core option. The Athlon 64 X2 3800+ may cost $87 more than its Intel counterpart, the Pentium D 820, but the AMD chip is a much better performer. It also uses considerably less power.'"
The best price/performance deal is the $146 AMD 3000 chip.
It is an amazing little bugger that can git er done with ease but does not cost and arm + leg.
If you "get" pointers add me as a friend (116)!
Q: Dr. Hurd, I am tired of reading you bash religious conservatives! You defend the "right" to abortion and gay marriage, for example. You care about the freedom of abortionists and homosexuals. What about freedom of religion?
A: There is not, and there should not be, any such thing as "freedom of religion." Think about what this simple little phrase implies. It implies that people have freedom not merely to privately practice whatever religion (if any) they choose; it implies that people have the "right" to impose the practice of religion on other people. Freedom OF religion is a blank check for religion being free to do what it pleases. The ultimate manifestation of this idea, that today we know all too well, is that of the militant Islamics.
There is no "freedom of religion" to do what it pleases. The only type of freedom relevant to religion is the right to practice religion privately, without imposing force on other people. If a religious person believes a fetus is a human life, then it's his right to believe so and to try and persuade others of this religious belief; the line stops, however, at forcing women to bring unwanted pregnancies to term. If a religious person believes that only religiously sanctioned marriage is a valid context for sexual relationships, then it's likewise his right to believe this; but again, the line stops at arresting or otherwise using the force of government to condemn non-religiously supported relationships (between consenting adults, of course) as immoral. If a religious person wants to put a nativity scene on his lawn, or at his church, for Christmas, this is certainly his right; the line stops, however, at requiring tax-supported courthouses and other government agencies to host religious scenes in the name of state-sponsored religion.
During the Clinton years, I strenuously argued against the growing statism I saw coming from the secular left. The Clintons tried to nationalize health care, turn doctors into slaves and to legally prosecute people like Bill Gates simply for being gloriously successful. Now, during the George W. Bush years, I proudly and boisterously oppose the religious statism of the right and Zonk. Through it all, I hope people come to understand that the only way to restore full freedom to this society--and keep the great number of freedoms we still enjoy--is to oppose statism of every variety. I want church and state to be separate for the same reasons I want economics and state to be separate. Stay out of our wallets; and stay out of our wombs, our bedrooms and our personal lives.
It costs almost $100 more and is faster? What are the odds...
The best argument against democracy is a five-minute conversation with the average voter.
- Winston Churchill
AMD
Is not gay
Intel
You go to hell
AMD
Your price is right
Intel
Puke, I might
AMD
I love you dear
Intel
You can kiss my rear.
I am teh poet.
Or, put another way, the bottom-of-the-line AMD 3800+ is less than 1/3rd the price of the top-of-the-line Pentium 840 EE ($328 vs. $999), yet it still beats it in most of the benchmarks.
Too bad they didn't compare the Pentium D 830 in the benchmarks - this is closer in price to the AMD 3800+
HIV Crosses Species Barrier... into Muppets
for $87 more i can buy a chip which is not only faster, but does not have DRM. nice!
VLC FOR MAC IS DYING! IF YOU DEVELOP, PLEASE SAVE IT!!
Coral Cache Link of the entire article in Print View
Why is it so hot? Where am I going? What am I doing in this handbasket?
Isn't this like the fourth time we've seen a Xeon-vs-AMD benchmark on the front page? It's old news.
The problem with the Xeons is they're totally throttled. The Xeon was like a V-6 engine under a VW carburetor; the dual-core Xeon is like a big-block V8 under the same carburetor.
The AMDs have better access to RAM and better (independent) cross-CPU communication. The dual-core Xeons were clearly rushed to market to answer AMD's offering, before Intel could get their own memory-access ducks in a row.
The AMD processor certainly provides a bit better performance for the price. Unfortunately, it seems that AMD still has several stability issues to solve... Perhaps this explains why major companies still prefer using Intel processors for building robust servers.
Won't somebody please make a commodity Power PC box I can run Linux on?
No folly is more costly than the folly of intolerant idealism. - Winston Churchill
Dual core shoot-out: Intel versus AMD
That is why I told everyone the best chip overall based on VALUE.
5 5&cid=13952965
You can get a brand new chip that is almost as fast as any other chip in the world, but at the PERFECT sweet spot in terms of price/performance.
Information here in my first post above that ironically 1 person modded off-topic in a thread about the best consumer processors: http://hardware.slashdot.org/comments.pl?sid=1673
If you "get" pointers add me as a friend (116)!
zdnet is usually fairly good, but not this time.
This doesn't really surprise me, AMD has outdone Intel in most (nearly all) benchmarks in the past, this is just continuing the trend, it's cool, though, to know AMD keeps beating out the deeper pockets of Intel.
I like suggestions, but I don't like contributing towards them.
A kernel compiled for a single CPU is faster than a kernel compiled for multipe CPU's, even when you only have one CPU. This is why OpenBSD has two kernels: 1) one cpu and 2) multpiple CPU's. The main developer of DragonBSD said that his preference is single CPU, performance wise (I'll leave that as a Google exercise).
- Cheap
- Fast
- Low power consumption
Pick any two.On se Internetz nobody noes your German.
(If you can afford it) I have found Itaniums to be very QUALITY platforms.
;)
They basically got as fed up as you with archaic x86 instructions and set out to make a good new platform.
Engineering-wise Itanium is the bee's knees.
Unfortunately they ARE expensive, so you probably have to be a full-time geek like me to afford one
If you "get" pointers add me as a friend (116)!
Hence, there's still a good reason to buy Intel instead of AMD.
The owls are not what they seem
Like most people, I will wait it out until the dual-core chips / products are stable and less expensive.
Not everyone is playing Quake 4 and Half-Life 2 on a daily basis.
He who knows best knows how little he knows. - Thomas Jefferson
This isn't a opteron-xeon review. Its a Athlon dual core vs. the new intel dual core (i.e. not the P4 line).
I mean hell it says it in the summary.
Remember when all AMD's were inheritly less then thier Intel counter parts? Times have changed indeed.
- w00t?
So let me get this straight. The AMD chip is more expensive and uses less power than the Intel chip? Isn't it usually the other way around?
I have used both GCC 2.x, GCC 3.x, and Intel's compiler on Itanium with much success, in addition to running and building all my Java code using 1.4.x.
If you "get" pointers add me as a friend (116)!
Ahh the AMD wins overall in performance but can it cook make me a sunny side up egg as fast as the intel :P
GL HF!
I read somewhere recently that 'more watts used' = 'more powerful'
Ignore anything I said above, I actually agree with everything you believe - mod accordingly.
Oh and no need to mention which kernels and OSs would be ideal; we already know about the answer to that.
"The Intel vs AMD battle of the benchmarks continues."
AMD has pretty much trounced Intel performance at every desktop and server pricepoint for the last 2 years at least, so who cares anymore? Even Dell has started carrying AMD CPU parts:
http://tinyurl.com/c57po
Dell is pretty much singlehandedly holding up Intel on the desktop, as they can drive the overall system price down on volume despite the higher-priced parts.
If their little Israel division hadnt come up with their M chips they'd even be worse off.
Sorry... but anyone who says "git er done" should be clubbed like a baby seal.
You are all a bunch of idots.
There are few of us in the IT/geek field, so I strive to stand out and attract fellow men into the profession ;)
If you "get" pointers add me as a friend (116)!
The ultimate multi-core processor technology is VLIW (or EPIC as INTEL calls it). The cores are broken up into lots of tiny pieces, instructions are distributed through various pipes and run through whatever is available in parallel. The Itanium processor is Intel's EPIC problem child. Too complex, too much heat. Maybe it is just a bit too early for this technology. I think Intel could try to start a "mobile" Itanium project. They were quite successful with their Pentium M. Maybe that will give Intel an advantage. ;-)
Or, Intel designs a dual-Alpha processor to beat AMD, but that sounds not like Intel, does it? Someone at AMD who might like the idea?
Your PC may have Intel inside, but did you know that Intel's fabs have VMS and Alpha inside?
AMD-optimized compilers are available from PathScale and AMD, but AMD processors will run Intel-optimized code pretty well anyway.
FEAR doesn't take advantage of it according to FiringSquad.
Ant(Dude) @ Quality Foraged Links (AQFL.net) & The Ant Farm (antfarm.ma.cx / antfarm.home.dhs.org).
And you should really stay in shipping where you belong with that kind of language. :)
That's why you install two hard drives to match the dual cores - then you run the antivirus/spyware scanner on the one that you're not using - i.e. the one without your OS and programs.
The same summary says both Athlon 3200 and 3800, which one is it?
The Pentium 4 is weak. The Pentium D isn't any better.
Where are the Pentium-M based desktop chips?! They are proven to be faster than AMD's chips.
It Intel just toying with us for fun?
They merely updated the p3, glueing other stuff together that already existed.
Call me old school, but back in the day when 283, 386, 486, Pentium, P2, and P3, I always praised Intel for their products over AMD in regards to performance and stability. Of course, nothing lasts forever. Which is really sad when you think about it. Nothing hurts an Intel Fanboy like myself then to see lackluster performace and innovation from the the very company that started the x86 momentum in the industry.
Times have changed. It's time I started drop my "trust" and "faith" and start going with AMD this time around. Clearly, they are the leaders in innovation this time.
Life is not for the lazy.
By my calculations, the power difference between the Intel and AMD will make up the difference in the chip prices in about a month of continuous operation, at lease for Seattle electric rates (~$0.06/kWh)...
Okay. According to this page, at full-tilt the Pentium D 820 consumes 130.6W, while this page says the Athlon 64 X2 3800+ consumes 89W. So, how long would the Opteron have to run at full blast to make up the difference in cost of $87? Last month I paid $0.078 per kilowatt-hour. This seems to be reasonably average for the United States. 130.6W - 89W = 41.6W difference between the two. Some back of the Google-calculator math reveals: (US$ 87) / (41.6 W * ((US$ 0.078) / (kW * Hr))) = 3.05871582 years. A not-insignifigant amount of time. If you're in an area where electricity is more expensive like New York or California, the amount of time is even less!
Feel free to correct my math!
"Yeah, we tried selling AMD products, but customers didn't want them!"
HTH. HAND.
Belief is the currency of delusion.
I have a question that has perplexed me for years now (since I built my AMD computer in 2001). If AMD consistently offers better CPUs for a lower price, why do they lag behind so much in sales? You would think at some point capitalism would kick in, people would leave Intel, and AMD would start getting a better market share. So, what's the deal?
I just ordered a new terminal server yesterday, and decided to go with this exact CPU... the AMD Athlon64 X2 3800+. This gives me satisfaction knowing I made the right choice in choosing to go with AMD (as I have for years now) instead of Intel.
Also, it appears that this particular CPU is VERY overclockable. People have been able to overclock from the stock 2.0Ghz to at least 2.4Ghz with stock cooling, and to at least a very impressive 2.8Ghz with minimal cooling upgrades.
Most of this information has been gleaned from a thread on Anandtech which I perused just yesterday. The thread/forum appears to be down at the moment, but Googling up "X2 3800+ Overclock" should bring you results if you wish to look them up yourself...
Google for "AMD Intel Antitrust" and see what you find. Basically Intel has allegedly been maintaining its monopoly via unorthodox (and in some cases plain illegal) means.
People like Dell are 100% Intel because if they sell even 1 AMD chip they will lose millions in back-hander "Advertising Funds" that Intel ply's them with.
The other reason they haven't been so popular in the data centre is that there has been a dirth of quality enterprise-level chipsets. The 4 and 8 Ways that Sun and IBM currently sell should sort that out thought.
The other problem is that the ignorant public at large still think AMD chips run hot and guzzle power when it's the inverse that is actuall true, Athlon/Turion/Opterons run cool whereas Pentium 4 and Xeon's are hotter than hell and suck up juice like a souped-up SUV.
I am NaN
Or produce some evidence for your blather. By all accounts, the AMD's don't just run rock stable, they run rock stable while seriously overclocked! You can't do that to Intel, they already are essentially max-overclocked, that's why the processor makes 250 Watts of heat. Yeah, go buy that. Try to keep it cool enough to run stable. I'm an otherwise nice person, but you've pissed me off enough to wish this on you!
Let's see, the AMD 3800+ has less than half the power consumption of the Intel EE, costs less than a third as much and beats it in practically every benchmark. That's the Cliff Notes, you illiterate moron.
I just bought the Pentium D 830 over the AMD whatever because the AMD-driven ultimate machine ends up not harnessing the extra speed. Considering that I get to use 8gbof DDR2 at 1044mghz and Intel Engineering instead of 4gb of ddr 400 as an intricate part of the dual core's ultimate performance building an Inel machine was a no-brainer. It's not the speed of the fastest part, but the slowest link in the chain that matters.
Relax, aren't you lucky that it is only my Opinion?
You mean it wouldn't perform well on real-world floating point applications then? Like SPEC fp 2000?
Spec fp 2000 results
Oh, I dunno, I think those AMD results look pretty good...
Yeah I just wanna give mad props to my registers for makin this thing HAPPEN
XMM0/MM0/EAX/AX/AL and
XMM1/MM1/ECX/CX/CL and
XMM2/MM2/EDX/DX/DL... I LOVE YOU GIRL...
and I can't forget
XMM3/MM3/EBX/BX/BL...
You are all wonderful, thank you again!
Er... shoot out?
We have secretly replaced these Slashdot mods' sense of humor with a rusty nail. Let's see if they notice!!
Waitaminute... The AMDs use less power and are faster than the Intels? And they're cheaper? Would somebody please explain to me again why Apple is moving to Intel?
If the reviewer wants to focus on value, it would be helpful to present each benchmark result as a scatter plot with price on one axis and performance on the other. Another set of scatter plots could present power against performance.
That would be more helpful than pretending that price and power trade-offs are the same for every reader. $87 means more to some people than to others.
Unless I've been under a rock for a century, I haven't heard of an AMD Athlon 64 X2 3200+..
AMD comes out on top quite rightfully but actually neither of these processors offers good value for, perhaps, the majority of all computer buyers. A great deal of what folks do - word processing, surfing, email, etc - can be done very well on a p3, a Mac Mini or even a Via Epia combo. The trend to bigger is better has simply landed people with behemoth-sized machines that are expensive to buy and run and messy to maintain.
It's also allowed free rein to OS bloat. And 1001 WinDel reviewers who'll gladly tell us that we really must have that 5-litre SUV to run the kids a couple of miles to school. That said, if you do need this kind of power then imho AMD's current chips offer a superb solution, but it's not for everyone.
Las qué passoun
tournoun pas maï
That is sound reasoning, in theory. Your memory bus may be faster, but the 2 cores still have to share the same pipeline to the memory with the Intel chips, severely bottlenecking performance, whereas with the AMD chips, each core has its own direct channel to the memory, meaning no fighting for time in the fast lane...this is where the AMD chips shine.
I would be willing to bet that your system with its 8GB of 1044MHz memory would still get trounced by a similar AMD system with its 4GB of 400MHz memory because that extra speed and memory is close to worthless if the cores are continuously fighting over it and cannot get a steady stream of access.
With all this new hardware, the case is 10deg cooler than when I had a P4 in there, off the same 500W power supply. I was still buying P4s when all my buddies had screaming AMD boxes, and I could not keep up in either Battlefield 2 or just recently in Call of Duty 2. I wasted a lot of time (and watts) sticking with Intel trying to cool their CPUs, but never again unless they take back the power/performance/price curve from AMD.
I aint holding my breath for that to happen.
Totally OT, but this is a site for geeks, and geeks like facts, figures and statistics... so... VW were actually the first manufacturer to utilise EFI and have, in fact, had EFI in cars (thereby doing away with the carbie) for over 30 years now...
A link to the Type 3 tells us that "Originally a dual carburetted engine, the Type 3 engine was modified in 1968 to include fuel injection, reputedly the first mass production consumer car with such a feature."
Anyway...
Specialist Mac support for creative pros, Melbourne
I thought Intel was going to be the performance/watt leader.
I thought this was Steve Jobs's big justification for moving to Intel.
I thought when you're hugely and provably wrong that people quit listening to you and buy better products.
Or did I slip into The Twilight Zone again without realizing it?
"It's the height of ridiculousness to say for those 9 lines you get hundreds of millions."
ASUS A8N-SLI Premium Socket 939 NVIDIA nForce4 SLI ATX AMD Motherboard - Retail $165.00 Item #: N82E16813131540
I thought it was a lot for a motherboard, but when I saw that the damn board could do everything, I had to have it. Newegg sells refurbs of the board for $110.00, but you have to be an idiot to buy a refurb motherboard.
Sometimes they're faster.
How can this be?
Context switching between threads expensive in terms of cycles on a microprocessor. A second processor can cut down immensely on context switching - or even virtually eliminate it when only two threads are active.
"It's the height of ridiculousness to say for those 9 lines you get hundreds of millions."
Now with the newest Nvidia drivers, you dont even have to have matching cards anymore. Though I have matching 7800GTXs, I could just as easily pop one of them into another machine, and still SLI with a 6800 or even a 6600GTX board. The newest games at 1600X1200 with full options on almost require SLI to play right. If you saw what Battlefield 2 looks like on my new rig, with everything turned up and NO lag, you would eat those words. Yeah, I got cash for nice game rigs, but you dont have to be rich to reap the benefits of SLI.
The thing that bothers me about all these reviews is they fail to mention that the Intel processors need (more expensive) DDR2 memory versus DDR for AMD. If one is going to compare prices of the processors, the cost of the faster memory required by should be included in the price of their processor. Also note, that when AMD comes out with Socket M2 processors, which support DDR2, then they should benchmark even faster.
Steve Jobs wouldn't lie, would he?
Best Buy can have you arrested
You all might like to know that the AMD systems kick Intel's ass when crunching the ClimatePrediction.net models.
Interestingly, the project at first used an Intel Fortran compiler which switches-off SSE instructions when running on AMD processors, and we are still living with those binaries in most areas of the project. Yet AMD still compares favorably!
Thanks Intel for costing a certain scientific research project valuable processing time. It may be the last time that most of us participating in this (widespread) distributed computing effort will choose Intel processors or compilers. Way to go.
Dell was offering the Demension 9100 system dual core 2.8ghz 820 and 24 inch 2405 widescreen for 1199 (after 100 dollar rebate). They offer the deal every few months, so basically, its buy a 24inch LCD and get a free dual core PC. Awesome deal...
I don't think it's so much that Intel's chips are throttled. The X2 3800+ uses PowerNow, and so it too throttles back and drops voltage down when idle.
I believe the real story goes back to earlier architecture choices Intel and AMD made with the P4 and original Athlon chips.
AMD decided for SMP ops it would be better to design the link between CPU and chipset with a star topology, so each CPU gets its own comm lines.
Intel chose to stick to the simpler bus topology for SMP ops, which reduced motherboard complexity and simplified chip design.
This still applies into the dual-core arena, although everything is moved on die. Intel's cores talk to the system on a bus, meaning that under load bus contention creates a bottleneck as each core tries to move data around. On the AMD platform, each core can talk independently, and contention issues are minimized.
Worse still, some Intel chip designs examined on tomshardware show they tom's calls: double-core, where two separate cpu cores are mounted on the same chip carrier, rather than having two cores as one die. This means that cheaper CPUs can be made because the dies are smaller, but each die much be matched so that they'll run well together under a common heat spreader. And again, their sharing data lines.
I'm too lazy to break out reference URL's for your examination, so I may have some factual errors, but I belive the general idea here is accurate.
USNG: 14TPU4605
Now, if YOU are poor, then all you said would be true, if you were refering to what to do with YOUR finances. Just thought I would clear that up.
Cool. The PC BIOS was really starting to look old. I guess people with cars that use OpenFirmware won't be able to laugh so much anymore...
I am TheRaven on Soylent News
...but what about motherboards? Last time I paid attention to Intel's dual core offerings, they would only work on boards equipped with the 9[4|5]5 chipsets. In the UK at least, the cheapest compatible board I can find from my fave hardware retailer costs £75, with the cheapest dual-core compatible AMD 939 board clocking in at under £50.
Bearing in mind that the 3800 X2 costs £230 and the P4D 820 costs £170 from the same store, for only £35 extra I get a duallie that'll thrash the P4D 820 and 830, plus using less power to boot, so within a few months it'll have paid for itself (if, like me, you count doing work faster as payment anyway - otherwise you'll have to wait a few months more for the leccy bills to add up).
My beef with these atricles and so many like them? The pseudo-TCO doesn't take the cost of the other hardware into account. A CPU is pretty useless without a compatible motherboard, no?
Moderation Total: -1 Troll, +3 Goat
It has an 89W cooling solution on it, true. But so does the 4200+. Starting at the 4400+, the chips come with a 110W cooling solution, you never hear AMD fans trumpeting that!. But anyway, if the 4200+ takes 89W, tops, then the 3800+, which runs only 91% as fast and is the same die must use even less.
http://lkml.org/lkml/2005/8/20/95
A friend of mine had a type 3 fastback with EFI. I can still remember the time we made a stop on a reasonable length trip and had to catch a train the rest of the way because the EFI decided it didn't want to work for the rest of the day. He spent a fortune on that car - and in the end the best thing he did was convert it to a carburetted engine.
And yes everyone should buy AMD...
dnuof eruc rof aixelsid
x86 is ugly and bad and nasty, x86-64 helps with more registers but it is still one ugly ISA. But the key is compatibility is maintained through microOps, risc like instructtions that the x86 instructions are translated into. Another big key is they have been doing the microOp thing for a long time and have built in some nifty caches. Also they allow the performance of many x86 instructions to suffer while finding the instructions that can basically have a simpler risc equivilent. There is some overhead to this but not much, what is funny is that in the mess that is x86 there is most what is close to a risc instruction set. The best thing about x86 is years of compiler optimization and new instructions like sse that allow x86 to offer perfomance close to modern risc systems. The engineers at amd saw this firrst and realized that they could make an alpha like processor with modern features, but there is a penalty.
HEMI means how the cylinder head is made or looks like Hemisphirical in this case.
Like a Harley shovel head looks like a shovel if you look at the cylinder head after you remove it and flip it over the cavaty looks like a shovel.
chips as accessories: http://accessories.us.dell.com/sna/productlisting. aspx?c=us&l=en&cs=04&category_id=6197&first=true&m nf=116&k=
even though, why would anyone buy anything from Dell?
I'll second that, and to add to it Opteron systems use NUMA, which have even bigger pipes and quicker access..
Oh, ya.. and my dads car is faster than your dads too....
once more into the breach
You're confusing "throttle" as in power saving and "throttle" as in can't pump enough data to the rest of the system. In essense, you are both arguing the same thing. :)
Ah. No. Not really.
The whole point of DMA (most modern HDs) is that the HD interacts with the chipset and not the CPU. Thus Direct Memory Access (DMA) vs. Programmed Input Output (PIO) where the HD interupts the CPU to handle IO requests.
Reality isn't quite as clear cut, but for the user's view on things. You are wrong.
Who will guard the guards?
Unpossible. There was only one bank/group of RAM. One bank of RAM means one channel to RAM, that means only one processor can access the RAM at the same time. So the busses have to be dependent on each other (stay out of each other's way).
I'm gonna have to go with the grandparent. There were few positive attributes to the Athlons at that time other than them not using RDRAM.
I just retired my Athlon XP about two months ago. I'm glad to be rid of it. They ran hot and were quite power hungry. And they had just followed the first Athlons (pre-XP) which were far worse. Between their massive heat output and the wonky heatsink attachment system, many Athlons were overheated or cracked and destroyed (mine included, which is why I ended up with an XP shortly thereafter).
Also, as the grandparent posted, the Athlon still had a measly 266FSB when the Pentium 4 had an 800FSB. The performance difference in any app that used memory even moderately was very large.
It wasn't their brightest day, but most importantly, AMD was at least trying to produce what the customer wanted (unlike Intel, who was in bed with RAMBUS) and over time, that has put them where they are today. Strong innovation and a customer focus put AMD on top.
http://lkml.org/lkml/2005/8/20/95
Why exactly are we posting stuff that we knew months ago? If you are smart you want the dual core Opteron 165 that runs the same MHz as the X2 3800+, but with a full 1MB cache per core, and comes in at $300.
ignorance is bliss. googlefiberatx.com
Comparing latency in cycles with CPU with different clockspeed is stupid so I sent this email to the author of the article:
>>>
Hello,
I noticed that in your article there is a table called latency where the various latency to access the RAM, cache are stored in cycle.
In my opinion, this is a mistake to have a table comparing latency in cycles (cycles/s is probably a mistake too) between CPUs with differing clock speed: say I have a 2GHz CPU accessing its cache in 10cycles and a 4GHz CPU accessing its cache in 15cycles, which CPU has the cache with the lower latency?
The 4GHz CPU of course because 15cycles * 1/(4*10^9)~= 3ns of latency
whereas the 2GHz CPU has 10cycles * 1/(2*10^9) = 5ns of latency.
So please fix this mistake..
Some of you that don't want to get the AMD chip because you can't get DDR2 RAM for it. In Q2 of 2006, AMD will be putting out a new CPU interface called the Socket M2. Motherboards that use this interface can implement DDR2 RAM.
I'm starting to feel the itch to build a new PC. Waiting for M2 will help me hold off.
Democrats or Republicans. They are both taking us to the same place and they are not afraid of us anymore.
"For example, a hard disk defragmenter may be running in the background, leaving insufficient resources for a foreground application like a presentation." Sounds like just the thing to cure the woes of just about every self-described "tech savvy" sales idiot I've had to support. Now maybe we won't have to field calls that their "computer is broken" when "all" they're doing is running a defrag while presenting.
There's only access path to RAM. Two devices cannot access RAM at the same time. It's axiomatic.
You're saying something akin to "I have two telephone lines in my house, so I can carry on two conversations at once!". It isn't true. There's only one of you, so adding phone lines doesn't mean you suddenly can talk to more people at once (at least not efficiently).
Look at it this way. If a company does phone support and has only one employee and one phone line, how many customers can they serve at once? One. Now if they upgrade to 10 phone lines, how many customers can they serve at once? One. How much does the system throughput here go up when you added more parallel phone lines? Not at all. If things worked like you said, then companies who do phone support would only have one employee, and 1,000 phone lines!
You can have parallel 300 busses in your system which are used to conduct data to and from the RAM. But in the end all those busses are dependent on getting onto the RAM bus. So, if you have only one RAM subsystem (as these machines had), you all your busses leading to RAM are dependent on each other.
Some recent Athlon 64 systems have multiple RAM subsystems. These older Athlon systems did not.
http://lkml.org/lkml/2005/8/20/95
Yeah, one processor could talk to something else theoretically. But there's really nothing else to talk to. Back then, perhaps the graphics card. Nowadays, the graphics card does most of its own work. And most I/O goes over DMA and bypasses the CPU completely, besides that, DMA all goes in and out of RAM anyway. So you're still stuck talking about the RAM bandwidth.
When my post's title was "you mean to RAM", that you replied to (and kept the title), you simply could have said "Well, it's not necessarily to RAM" instead of "it doesn't have an integrated memory controller".
Anyway, I thought it was clear in the post you responded originally that the poster was referring to the poor performance of the Athlon systems of the time due to poor memory bandwidth, not because the CPU couldn't talk to the floppy disk controller fast enough. And even if these busses were operating independently at any given time, the memory bandwidth was still very poor. An Intel system with 800FSB and dual-channels could theoretically transfer 6.4GB/sec. The AMD with 266FSB and only one channel only could transfer 1.066GB/sec.
That was the performance problem, and parallel channels from the processors to the memory controller didn't do anything to alleviate it.
I have an additional question. If these two busses were truly independent, how would the processors keep their caches in sync with each other? I mean, if one processor is talking to the ethernet controller (for setup, data is DMAed) and the other tries to talk to RAM, how does the first find out what data the other processor is getting from RAM (or even that it is doing it) so that it can snoop and keep its caches valid? I mean, this kind of problem is what hurts Intel right now in their MP systems, there's no reason to think it didn't limit the ability of Athlons to access memory concurrently back when AMD used that architecture.
Do you know how AMD got around this?
http://lkml.org/lkml/2005/8/20/95
If you are truly a man, then you should have a penis. So where is it? I'm looking directly at your genital region, but I do not see your penis!
Cyric Zndovzny at your service.