Xeon64 or Opteron

[bobbyk]

It might, but it also might be engineering samples. Of course, intel wanted to be sure that in doing this it would be compatible with the os. Microsoft said they were only making one and amd got there first. It probably didn't run at the speeds that we will see upon release; however, it should be noted that the people that make an instruction set first would be able to figure out how to run it better an faster than someone who follows. Look at p4 sse2 performance.

 

[SickBeast]

Originally posted by: bobbyk
It might, but it also might be engineering samples. Of course, intel wanted to be sure that in doing this it would be compatible with the os. Microsoft said they were only making one and amd got there first. It probably didn't run at the speeds that we will see upon release; however, it should be noted that the people that make an instruction set first would be able to figure out how to run it better an faster than someone who follows. Look at p4 sse2 performance.

What I gathered from the article that I read was that AMD's 64-bit implementation was more efficient than that of intel's.

 

[bobbyk]

good to hear. I thought it might be this way, but it was also important to point out that the xeons were probably not full speed. Microsoft must have had them for a while now. I think amd will stay on top for a long time. They have already said, if needed they can do the x52 series and the fx-55. My guess would be they get to wait on intel now and look to future improvements for the processor.

 

[Nemesis2038]

I do belive the longer pipelines are a problem for the CPU. Thats why I would rather see a shrunken P4 over the prescott with longer pipelines. I dont think we will see a benefit to the longer pipelines until the prescott processor reaches around 4Ghz where it can really start to utilize a longer pipe. In the mean time I think its a waste for intel. They have a performance gap in the 3.4-4Gig or higher range until that pipe can be used effectively. A better method might have been to just shrink the P4 than to have designed prescott.

Sorry if I misquoted you on the on the SSE3 bobbyk. Someone is saying it will arrive in the 90nm. Which is good despite no one is really using it except the benchmark utilities.

Also good point on the integrated memory controller. But nothing is really known about the Xeon64. If its a reverse engineered AMD 64 then AMD could be in for some competition however if its emulation 64 then AMD will easily smoke it. Either way its a warning shot off the bow of AMD to keep things moving.

I think of 2 things sort of off track but on.
1. Existing Prescott 64 must be really bad or not compatible or broken if Intel wont turn it on. Or they wont turn it on until there is a 64 bit OS. I doubt it will ever be turned on. I think it may be broken or possibly the additional transistors may turn the heat up enough on the proc that it may cook itself. I doubt anyone knows the real answer to this we can only speculate.

2. Xeon64 would make Intel look really bad if they release it and its a dawg. I think Intel would want to show they are in control to some degree. Its a toss up. But with talk of it being released at 3.6 is pretty impressive but if its just a fixed prescott with 64 bit emulation then go AMD. I guess we will know on monday.

Nice link ALIEN3001 I watched a little and will educate myself a little futher tommorow morning.

This is much better communication that earlier today. Your all restoring my faith in my Athlon64 purchase. Thanks Again Everyone.

I still would like AMD to create SSE4 or their version of it and whats needed.

 

[bobbyk]

SSE4 wouldn't really be worth it. When we have the R480 and the NV(whatever it is), they will be enough to handle those instructions. Plus, they are optimized for it which means they run them faster.

2nd. I think Intel was forced to do 64 bit. Currently Amd is spinning that everyway they can. They have an advantage even if it doesn't work yet and regular people won't know it. AMD can cite actual performance increases because they exist. No one disputes that. THe thing is if for some reason AMD got a marketing lead intel would be hurt terribly. I am not sure that they could reverse engineer the a64 because it is full of patents and copy protection stuff. (even with the shady agreement, i think that some of amd's later patents are protected from intel) It is also difficult to do something like that with the xeon architecture. I don't think it would work as well as it does for amd.

It is kinda like the hyperthreading argument. Intel pushes that for all its worth. The actual performance benefit is about 5 or 6% in common applications. In encoding it is huge but that is a different story. Hyperthreading if even possible wouldn't work as well for amd because the architecture is different. All it really does is keep the cpu working if a pipeline stall occurs. The a64 only has a what 16 stage pipeline if i remember correct. It doesn't stall as much and it is shorter so it doesn't hurt the amd cpus as much as it does the prescott with the 31 stage pipeline.

The difference in the architectures makes copying each other a little bit difficult.

 

[imported_joelberg]

Wow, the Xeon64 IS just a Prescott with 64 turned on! thats why it sucks! We all have read that the Prescott's 64-bitness is emulated, how COULD it be faster than opteron? Maybe in 32-bit the knew Xeon will be able to compete....at 3.6 Ghz lol. It won't compete any more than Prescott does, and that's not much unless you are talking about 3.4 and 3.6 Ghz processors.

 

[bobbyk]

we don't know yet, actually. The prescott and xeon nacona cores might be very different. IN the past intel has used a largely p4 based architecture for their xeons, but this might change that. I do not think that is the likely case, but it is possible.

We also don't know if it sucks yet.

I might get flamed for this, but in theory prescott isn't that bad. Once better revisions come out it will offer at least the same performance as northwood. Northwood is actually an amazing core that intel has taken to heights far beyond what it was called to do.

If amd can introduce a new core revision with better ddr support (more optimizations), sse3 (maybe), and be able to ramp clock speeds without doing 90nm, they will have a fairly secure future for the rest of the year and well into the next one. If amd somehow is the only company not having problems with 90nm it would be very good for them. I think one of the reasons that they are not having as many problems with it is because they started much slower than intel. Intel now needs clock speed. They knew the pipeline of prescott would hurt them majorly until it gets above 4 GHz. The mfg. process they use creates very high heat though. Strained silicon is good, but it allows for much less stability and more leakage than the amd processors because of thinner layers in crucial areas of the chip. SOI is the opposite. It looks to reduce leakage by adding something else betweent the layers. This may be why amd isn't having the heat issues of prescott yet. THey probably are still coming.

The only wildcard in the whole thing is what is wrong with IBM. They have major 90nm problems. They use soi. I don't think amd is having the problems of ibm, but that doesn't make sense. They are chip design partners. the 90nm ibm fab is also the worlds best. East Fishkill is huge. It isn't working right, but when it does it can probably do 65 nm now. They (microsoft and ibm (xbox 2 chip)) said they had working 65 nm now. That isnt official i do not believe, but it is possible when you have the money that ibm does. Their problems are evidenced by apple's "new" water cooled furnace (G5). This just doesn't quite fit. Ibm has been doing 90nm since last year, why isn't it working now. (i realise that is kinda off topic, but ohh well)

 

[Cerb]

Originally posted by: Nemesis2038
This is my pure speculation and I have nothing to base it upon except general reading of Nocona information found around the web which isnt much to go on.

I think Intel may deliver a pretty darn good competitor to AMD Opteron on Monday. Maybe even leapfrog Opteron in performance. That is most applications. 32bit apps I think will stay in comparison to the P4 Prescott and Opteron comparisons but on the 64 bit level I think Intel may have something good.

From a lot of reading it will probably get the majority of the benchmarks and AMD will need to put out a faster Opteron or increase some cache to overtake it.

The benchmarks look goo to me. 2.4 and 2.6 kernels love the Opteron.

I cant see any OS advantage since Microsoft will write the OS in a generic format to support both processors and will leave out any advantages of code that the other processor does not carry. We all know how good that Wintel Alliance is. If CPUID = AMD then Wait ELSE Continue.

MS got squarely behind x86-64, and even in 32-bit, Linux does a good job. MS just needs to make the OS NUMA-aware and compile it with their own compiler.

But I can see application performance leading to Intel in the future. You know they always come out with something like SSE64 which applications will use to enhance performance. Something AMD is proably too stupid to do with the Opteron when they were the only 64 bit x86 chip in town 3DNOW64 should have been in the AMD 939 Chips.

Bah. They did x86-64, and got NX support before Intel.

I made that enhancement area up but Intel's SSE has added some nice benefits to its CPU's. When AMD was the only 64 bit chip they should have started doing the same. Instead they have a plain jane X86 64 extension. Nothing enhanced about it. Simply making Intel AMD64 compliant wont be enough to keep Intel from passing up AMD opteron.

No, plenty enhanced. Look at the non-gaming benches for XP64 (gaming ones are still too limited by drivers).

AMD may take the lead with multiple CPU's but I suspect Intel will come out with a way to get past that too. Hyper transport does nothing unless you have 2 or more CPU's. I give the advantage early to AMD but favor Intel in the long run.

Wrong again. The P4's FSB is also being taken up by transfers from the RAM. The A64 can use the HT links separate from that, hence not technically having a FSB.

Benchmark programs I never trust since they are always skewed toward who donates the biggest check.

Games and ZD's stuff can usually be trusted.

Like how benchmarks show there is no possible way an Opteron can outperform a P4 but most applications and all games perform better on an Opteron.

Not all, but most. Those benchmarks, save for Aquamark, are using the actual game and playing a demo. You're benching it based on the actual game performance. it doesn't mean you'll get 80FPS in-game 0if the demo gets that, but if you run the same demo and get 60 FPS, you know that your actual game FPS are about 25% behind.

You would never know that by the benchmarking programs out there. Scammers. Give us lots of money and our next Benchmark will favor your CPU. But people fall for that as if a benchmark actually completed a task. Benchmarking programs unless based upon real world applications are useless to me.

Aside from making sure all is working to par, I agree. 3Dmark and Sandra only let you know that you're about where you should be, and don't have any erroneous performance settings.

I like my AMD chips but I can see Intel leapfrogging AMD and AMD not innovating enough to stay ahead. Intel just has that much more pull in the community.

But that appears to be erroding, actually. And how can you say using HT, going to onboard memory controllers, and making the K8 a NUMA system (there's a better word, but it isn't hitting me) is not innovation for x86? They suprised everyone with that, and have been able to keep on schedule with releasing speed grades, something they had real trouble doing with the Athlon XPs.

What might be a big surprise if Intel doesnt have a complete x86 64 extension and resorts to some sort of emulation type 64bit. However I dont think that will be the case.

Neither do I. AMD has too much to gain and Intel too much to lose.

In the end I hope pricing makes chips cheaper for whatever side of the fence you ride on. I ride both sides so it just means Im getting a cheaper CPU. Personally I dont care who makes the processor as long as it meets my needs and is cheap. I favor no one.

I favir AMD, but because they offer the best features at the best price for what I want. CnQ vs. Prescott 100+ watts.

Cant wait till monday. Hoping for some good benchmarks.

Whats everyone else think?

 

[Cerb]

Originally posted by: Nemesis2038
In short I dont think Intel would announce a 64 bit cpu unless it rocked. Coming out the gates slower than AMD's Opteron would make Intel look bad. I expect more from intel first x86 64 bit cpu. Surely because I am so far dissapointed with Prescott. Maybe the Dothan team needs to give the Precott team a lesson in CPU design. Just my thoughts.

I dunno...Willamette wasn't impressive, either.

 

[Cerb]

Originally posted by: Nemesis2038
Hypertrasport is really for those with Dual or more chips. its not innovation unless you plan on going that route. Its a bonus for AMD for future processors because it allows easy migration to dual core manufacturing.

Lets look at intel. SSE, SSE2, SSE3, HT(Hyper Threading).
AMD 3DNOW, 3DNOW+(I Think) Then its Copies of all Intel technology from here.

x86-64, NX bit, NUMA design, separating the FSB.

SSE is a great enhancement. These are items that actually do increase performance and are needed instructions. They had the perfect opportunity to expand 3DNOW and they blew it. All they have is an expanded X86 64 bit processor that has great FPU performance. Granted they shouldnt call the next generation 3D-Now because it brands AMD as a TOY or Gaming processor and not a buisness processor.

I dont consider branch prediction a feature but basic chip enhancement. Neither do I consider pipeline depth as thats just knowing how to load your processor with the right balance so it always has data to work with.

If AMD planned on being the leader it should have done some innovating beyond X86 extensions. Whats next 32 registers instead of 16? Thats not innovation its expansion.

For all intensive purposes the following should have been or be planned to be added to future processors.
DIVX ENCODING/DECODING OR INSTRUCTIONS TO FUTHER ENHANCE THE PROCESS.
WMV9 ENCODING/DECODING OR INSTRUCTIONS TO FUTHER ENHANCE THE PROCESS.
MAYBE SOME REVISED 3D NOW CALLED SOMETHING ELSE.
MAYBE EXPAND ON SSE. I BET INTEL IS ALREADY GOT PLANNED SSE4 or SSE64.

Or not.
Let's see...Intel is going with GHz, which increases SIMD thruput by quite a bit. AMD wants clock speeds to be very balanced, so they don't suck at it, but let's face it, GHz matter for dumb uses. Why enhance video encoding when Intel has done the work, and all AMD needs to do is license it. I have a feeling getting real 3DNow! support showed them how much work it really is to get software people on your side w/o donations.

These should have been added to Opteron 939 instead its the same CPU with non registered dimm support. Thats really a lame goal and shows they arent thinking ahead or thinking how they can 1 up Intel.

Wrong. SSE3 is on the new A64/Opteron core.

They talk about being a leader and adding features for the future way people work then why the heck didnt SEMPERON have any enhancements like this??? DIVX Code/MPEG4 is overdue in Processors!!! Heck even MPEG2 is overdue in processors. Simple things like this seperate a leader from a follower.

Because the Sempron is cheap, and they want to differentiate it so that you want the A64. I do think it was a mistake not to have NX bit support, though. That would have been awesome.

I fear AMD may miss the boat. Intel will release a 64 bit CPU then throw in SSE64 the software companies will capitalize on it and AMD will be following INTEL into what they started a year earlier.

So? AMD has been competitive in gaming and general use since the start of the K7. So what if the P4 encodes faster? Yes, that's right, encodes video and audio a little bit faster. SO WHAT? How many people does that matter to, versus photo editing and games? Not many. Oh, and HPC folks, many of whom are gaga over the Opteron.

I must give AMD 1 Item of innovation. A No Execute instruction that supposed to do some enhanced anti-virus/security function. BIG WHOOP!!!

You dont see this with the 2 big video card manufacturers. They are both adding thier own sets of functionality into their cards hoping to create something like Intel's SSE. ATI has their NMAP and NVIDIA has their funky programmable dohickey I cant recall the name. Both trying new things they hope will be something they force their competitor to have to liscence giving them the 1 Up. AMD should do the same instead of WE BUILT A 64 BIT X86 CPU!!! WOOHOO. Then do nothing more than try to make it faster and faster.

They are, though. Look at the benchmarks. They aren't adding extra instructions, but they sure as hell are doing something right, because they have a winner on their hands.

Video Cards is a game of Gladiators battling to beat each other silly.

But we care more about DirectX support than any extra BS.

Processors are David vs Goliath where once David knock out Goliath it thinks it will stay down. Instead Goliath will get up and Pummel David because he sat back victorious over a small win.

No. AMD is getting a serious foothold in servers. That is higher margin market, and once they get their teeth in, they are there for a long time. They pulled it off wonderfully, and the desktop derivative is pretty darn neat, too. And Intel is just now talking about doing something like Cool & Quiet, which is the single feature in my mind that makes the Athlon64 superior than the P4.

I give this troll a 3/10.

 

[Cerb]

Originally posted by: Nemesis2038
Not so 3DNOW is a rip off of MMX. MMX existed before 3D NOW.

And the 32-bit x86 instructions were there before MMX. Your point? They are all just patches to the system.

Also some of you need to read my first sentence in the beginning of the thread.
This is my pure speculation and I have nothing to base it upon except general reading of Nocona information found around the web which isnt much to go on.

This isnt to start a war of any sort. Get over it. In fact a lot of you need to stop being on one side of the fence or another. You should buy the best processor for the price wether it be from Intel or AMD. AMD and Intel's main goal should be to sell the best processor for the best price. I am happy the two of them are fighting it out so that we as consumers get cheaper faster chips. I have no alliance or favor neither manufacturer. Neither should you. You should buy the best processor you can afford not a brand name because you like david or goliath. These are not biblical times. To win my dollars deliver me the best processor that does the job the best.

Yes my Opteron is still 2-3Times faster at specific applications than my P4. But we are talking about the Xeon64 not the P4. I truely embrace the 64 bit side as there are many advantages for the things I do with it that 32 bit simply cannot handle. I explained what I was doing and how I got my results. If you still dont understand it then your the same type person who cannot understand how you can record one channel while watching another type people in my young days working at radio shack.

Expansion is Invention but its not INNOVATION? I would like to see AMD INNOVATE more and Take a much more responsible lead in that arena. I dont think SSE shoved anything down anyone throat. Intel said here are some additional instructions which will give you a performance boost. Either use them or use the standard x86 instruction base. Its up to you.

Itanium is actually a brilliant CPU design. Hence it was a chance as it did not support backwards compatibility and was a radical change.

And it costs too much.

Had AMD opteron sucked we would all be moving toward Itanium at some point. The Itanium is Intel attempting to drop the things that hold back x86 architecture. Just like Microsoft would like to abandon 16 bit application support. Intel would like to abandon 8, 16, and 32bit support. But the X86 still contains these things which hold it back from being a bit faster. Xeons had 16 bit support removed and its allowed for a speed increase, Transistor count decrease, and thus heat and power decrease. This is smart. You cant blame Intel for the Itanium. I commend the Itanium team its just that AMD delivered a great 64bit x86 cpu.

Which has better bang/buck, too.

Multimedia is Key and processors will only be found in computers if they continue to leave out DIVX/MPEG2/4/WMV9 If the cpu expanded on these Items it would open the CPU up to different markets other than a box in the office. Most will understand this more as HDTV becomes more mainstream. I believe these are items not neccessary for other chips like the graphics card to handle. There are a lot of reasons to implement them into the CPU. Mainly so that Integrated motherboards which users dont need an NVIDIA or ATI card can use. The majority of PC sales are to not Include add on video cards. Also to expand the cpu beyond the PC market.

Yeah, I think IBM and Motorolla kinda ate that already. Oh, and those guys making PDAs using them ARM thingies.

I think Ardrid makes some good points. Thats what Forums are about exchanging information but a lot of you look at it as battlegrounds. What war are you trying to win? Why are you figting for Intel or AMD. They should be fighting for you and your dollars.

So intel has to adopt AMD 64 its no different than AMD Accepting MMX, MMX+, SSE, SSE2, SSE3. Yup I forgot MMX something else AMD had to implement because it was Innovative.

So let me state this because I have no bias. Both AMD and INTEL make great CPU's. Each does something better than the other. Much like I have seen the comparison of the Truck vs the Dragster and moving. You either want to move a lot in one shot or make a lot of fast little trips. You decide what CPU is best for you. But for gods sake quite telling people thier P4 or Amd Athlon sucks because it doesnt. If the CPU does what they hope to do then its great because its doing what is was meant to do. And 500FPS doesnt make you a better gamer if you arse is wiped up by a guy only playing at 60FPS.

I would like to see the following. AMD come up with some more fresh ideas to enhance the 64bit experience with some more multimedia integration.

My God man, you sound like a manager! The multimedia experience will get better with better software and faster hardware. Since the 386 and ISA SB16, the 'multimedia experience' has been the prime buzzword, and still is, but the most innovation we've gotten from it is surround sound in games and 3D games. Aside from that, it just looks fancier. Window dressing.

I would like Intel to recosider the Prescott and die shrink the P4 instead. I dont see deeper pipelines are whats needed for 4.x GHZ and above. Time will tell.

I think we all want Dothan desktop chips.

I would also like to see some lower powered fanless CPU's for Multimedia applications. Think guys.

Not gonna happen. You do it with Durons, but that's as far as it goes.

Some of you take this stuff too seriously that I wonder if you would even game with a guy who has a different CPU or opinion about a manufacturer. Thats really lame and your missing out on something called LIFE.

Yeah, life. That's that thing they have in movies, right? With girls?
:beer:

 

[imported_michaelpatrick33]

the a64's do not currently have sse3. You are aware that Intel cancelled the Tejas processor successor to the PIV and now they are going to try and dual core flame throwing prescotts? Do you think the shorter pipeline PIII mobile that is coming will use hypethreading. Hyperthreading was in response to super long pipelines. it will be interesting to see if intel even keeps the hyperthreading with the shorter pipeline processors coming (except for the dual core Prescott

 

[Zebo]

Yes, that's right, encodes video and audio a little bit faster. SO WHAT?
--------------
That's not true. AMD owns LAME, DVix AVI and many other video and audio encoding apps now. It's more like split depending what app one chooses. Certainly not the domination AMD has in games, office suites, FPU work, compiling and other real world test suites.

http://www.hardocp.com/images/articles/10860382713dbCdJgsnE_3_1.gif
http://www.hardocp.com/images/articles/10860382713dbCdJgsnE_3_2.gif

 

[Cerb]

Originally posted by: Nemesis2038
If an Itanium were to be able to run at 3Ghz it would smoke and Opteron. Unfortunatly for Intel at Itaniums current speed the Opteron can outrun it now. Like I said it was a gamble for Intel and unfortunatly AMD played a bigger hand. AMD had COST, UPGRADABILITY FROM EXITING, EVENTUALLY SPEED.

If the Itanium ran at 3GHz, it would also challenge Power. At 2GHz, it gives everything out there a run for the money in floating point and DB work. Just too expensive for what it does.

How many times in the past has AMD promised incredible things only to be delayed and not deliver top performance. I think 90% of the life of the K6 did that.

Thing is, we knew it was crap .

The Athon originally was off to a big win outpacing Intel to the 1Gig mark but fell quickly when they didnt keep up the pace. Even the Opteron was delayed nearly a year.

So was the Prescott, and most of us kinda expected such things. Corporate hype is strange.

Granted they are doing much better now but its still not enough. AMD is notrious for jumping ahead only to have Intel come back bigger stronger and faster. You have to run it paranoid that the shark wont catch you because once you miss a stroke Intel will be there to tear aleg off cutting your swimming speed down. Before AMD knows it Intel will tear that second leg off and your going no where. Then AMD will discover the motor and swim past the shark only to realize the shark has friends everywhere and despite your speed increase there is a shark every direction you turn.

Yes, but in both cases, we win due to competition, provided they keep switching off. This time Intel missed the mark and AMD hit the bullseye. Next time they at least need to be prepared to hit their mark, but for Intel to match it or beat it.

I wish AMD luck and continued success. Intel needs AMD to be succesfull or Intel is a monopoly. This AMD spurt is a good thing for Intel. Much like Linux is good for Microsoft without it Microsoft is a Monopoly. Co existance is good for everyone. Especially the Consumer.

I am also not looking for respect. I get respect from the wife, kids, and Co-Workers thats all I need and then some.

I think whatever is added to Prescott under the hood is the problem with Prescotts heat issues. I suspect if they die shrink the P4 they will get a cooler, cheaper, faster, chip. Especially since they wont enable the 64 bit extensions on Prescott. Then its a 32 bit CPU the extra in there is just baggage holding the chip back.

...or if they'd just get Dothan to 2.5GHz and put it on the socket 755. All of a sudden you have a chip with good power use. Extra speed taken from core tweaks and extra voltage to get it running all super on the desktop, but still not a space heater like the Prescott, and compettitive performance across the board against AMD.

In general 64 bit doesnt add much unless you get into enterprise applications and some of the higher end stuff like when I am doing HDTV or Decrypting. Compression helps but who in the general public does this. Intel was right but wrong. 64 bit is not needed but does add some functionality and improvement only a small percentage of people can appreciate.

...but it isn't the 64-bit part that does that. extra registers and such take care of that.

Gamers, Encoders will see a benefit. Those who compose e-mail and balance checkbooks will just waste money with getting a 64 bit cpu. unless its bill gates who needs 64bit to balance his checkbook.

...and those looking for the best balance of quiet and performance will benefit, as even without CnQ enabled, Athlon64s run pretty cool against the competiton.

You bring up a good point Cross Liscence adds a lot for the two. Im not sure when or if it runs out. Anyone else care to comment on this and how it effects innovations from both sides going forward?

 

[Zebo]

Originally posted by: Zebo
Yes, that's right, encodes video and audio a little bit faster. SO WHAT?
--------------
That's not true. AMD owns LAME, DVix AVI and many other video and audio encoding apps now. It's more like split depending what app one chooses. Certainly not the domination AMD has in games, office suites, FPU work, compiling and other real world test suites.

http://www.hardocp.com/images/articles/10860382713dbCdJgsnE_3_1.gif
http://www.hardocp.com/images/articles/10860382713dbCdJgsnE_3_2.gif

Here is another link debunking the myth that Intel is faster at encoding. Notice Oggenc audio, Windows Media Video 9, and Conopus Procorder AMD leads the pack. It's really a shame most review sites continue to rely on just a couple apps perpetuating Intel's superior encoding MYTH.

http://www.digit-life.com/articles2/roundupmobo/athlon-64-3400.html

 

[n0cmonkey]

Originally posted by: Nemesis2038
Multimedia is Key and processors will only be found in computers if they continue to leave out DIVX/MPEG2/4/WMV9 If the cpu expanded on these Items it would open the CPU up to different markets other than a box in the office. Most will understand this more as HDTV becomes more mainstream. I believe these are items not neccessary for other chips like the graphics card to handle. There are a lot of reasons to implement them into the CPU. Mainly so that Integrated motherboards which users dont need an NVIDIA or ATI card can use. The majority of PC sales are to not Include add on video cards. Also to expand the cpu beyond the PC market.

Leave it on the video card, where it belongs. Look at VIA's solutions. The mini-itx boards have decoders (and maybe an encoder) in the onboard video.

I would also like to see some lower powered fanless CPU's for Multimedia applications. Think guys.

VIA C3. I want a couple for encryption work. The onboard multimedia doesn't sound all that bad either.

 

[n0cmonkey]

Originally posted by: srg
Hi Nemesis2038

Would you consider Hyperthreading an innovation??

srg

It would be for the company that came up with it.

 

[glugglug]

As Intel's new CPU will exist only on paper, you can easily overclock it to 36GHz. Just erase the decimal and draw in a new one.

 

[n0cmonkey]

Originally posted by: Nemesis2038
I expect the Xeon64 to be even faster than Opteron because of the deeper pipelines and the 3.6Ghz possibility.

You're a numbers guy. I understand now.

Yes Hyperthreading is Innovation despite it having some positive and some negative effects.

SMT is innovation for whom?

Granted I think Dual Core on the existing socket is really cool and INNOVATIVE but there will still be a lot of people out there with no need to have dual cores on a single chip.

And there is a need for SMT? Oh, and dual core is not innovation for AMD. IBM and Sun did it first

Maybe they should also look to ATI and NVIDIA about putting a Graphics chip right on the CPU core as well. Why not? XBOX2 looks like it might do something along that line maybe not.

K.I.S.S.


I'm an AMD fanboy at the moment. AMD64 chips are the best for me. If Intel came out with a chip as good or better and supported it as well as AMD is supporting AMD64, then I'm there. But it won't happen.

 

[mjuarez]

In short I dont think Intel would announce a 64 bit cpu unless it rocked. Coming out the gates slower than AMD's Opteron would make Intel look bad. I expect more from intel first x86 64 bit cpu. Surely because I am so far dissapointed with Prescott. Maybe the Dothan team needs to give the Precott team a lesson in CPU design. Just my thoughts.

I dont' know why they launched Prescott then... even their own Northwood chips at the same frequency were faster. Not to even mention Athlon64.

I think Intel will launch a sucky 64-bit Xeon on Monday, just so it has something to compete against Opteron. Companies will continue to buy Opteron by the truckload, and mainly dismiss the new Xeon, since performance will be even worse than today's Xeons, it will dissipate more heat than Prescott, and it will be a paper launch.

Marcos

 

[n0cmonkey]

Originally posted by: mjuarez

In short I dont think Intel would announce a 64 bit cpu unless it rocked. Coming out the gates slower than AMD's Opteron would make Intel look bad. I expect more from intel first x86 64 bit cpu. Surely because I am so far dissapointed with Prescott. Maybe the Dothan team needs to give the Precott team a lesson in CPU design. Just my thoughts.

I dont' know why they launched Prescott then... even their own Northwood chips at the same frequency were faster. Not to even mention Athlon64.

And the p3 was faster than the p4 at the same clock. I think the entire point is the possibility of ramping up the mhz faster and more easily. Where AMD has chosen to efficiently use the mhz they have.

I think Intel will launch a sucky 64-bit Xeon on Monday, just so it has something to compete against Opteron. Companies will continue to buy Opteron by the truckload, and mainly dismiss the new Xeon, since performance will be even worse than today's Xeons, it will dissipate more heat than Prescott, and it will be a paper launch.

Marcos

I'm going to disagree with this. I think the 64bit xeon will be ok, not better and not much worse than previous xeons. Hopefully they'll bump up the FSB to compensate. Companies will buy the Xeons because that's what Dell will sell them.

 

[IvanAndreevich]

Nemesis2038
K9. Ever heard of that? Do you want me to speculate on it?

 

[sandorski]

3DNow was not a ripoff of MMX. It was similar, being a special instruction set, but its' instructions were focussed on entirely different things than MMXs'.

 

[LoganTeamX]

I don't know about this Nemesis guy. I heard about all this FUD from Ard over at forums.amd.com and I figured "hey, why don't I go give them some numbers?"

I did a couple of interesting benchmarks involving encoding. The results are a little... disturbing.

http://forums.amd.com/index.php?showtopic=17088&hl=ulead

The bottom line is that if Intel wants to make a respectable processors that meets and beats expectations, they'll make their next processor based on the Dothan architecture.

But they're pig-headed, and they won't. Because that would be too easy.

The fact remains that if you need a processor that doesn't choke more than Italy in the 1998 World Cup, doesn't need to be schizophrenic in order to get work done effectively, and can do scientific mathematical computations in a reasonable amount of time... you buy AMD.

If you're going to sit there and love NetBurst and pretend you can rip a DVD while playing Halo... well, I've got nothing else that can help you.

 

[Vee]

Originally posted by: mjuarez

In short I dont think Intel would announce a 64 bit cpu unless it rocked. Coming out the gates slower than AMD's Opteron would make Intel look bad. I expect more from intel first x86 64 bit cpu. Surely because I am so far dissapointed with Prescott. Maybe the Dothan team needs to give the Precott team a lesson in CPU design. Just my thoughts.

I dont' know why they launched Prescott then... even their own Northwood chips at the same frequency were faster. Not to even mention Athlon64.

I think Intel will launch a sucky 64-bit Xeon on Monday, just so it has something to compete against Opteron. Companies will continue to buy Opteron by the truckload, and mainly dismiss the new Xeon, since performance will be even worse than today's Xeons, it will dissipate more heat than Prescott, and it will be a paper launch.

Marcos

I think the Nocomo is primarily a stopgap solution for Dell. Dell is overun with corporate buyers demanding Opteron servers. And now Dell is starting to become a bit worried about that clandestine, illegal marketing agreement they have with Intel. The one by which they use their considerable market position to squeeze out AMD from the budget sector, with lousy Celerons, in return for cheap cpus and being a priority customer. Dell have started to make a lot of squeeky, desperate noises for a some time. They are losing their grip on some old customers. And Intel have to to do something fast.

Meanwhile, Intel do work on a real '86-64 cpu. But it will be a while yet.