Last message on previous page:
More likely K8 cause since Phenom is native quad-core and Kuma must be native dual-core, thus requiring a die redesign... I'm sure any sane AMD fanatic knows that dual-core is coming of age.. since AMD is so behind it wouldn't make sense to leap backwards... unless they wanted higher clocks...

Still I believe they are focusing on Shanghai and Deneb and shouldn't focus on Kuma...

uguv has me pegged.

I just see red when there is too much blue.

Mrs Bytch may have struck a chord with the recent Havock news too !!

The Havok news shouldn't strike any chords with anyone. ATI always work together with Havok and Intel. In fact ATI and Intel were the ones who were prepping CFX before AMD decided to buy ATI.

Either way its part of Intel so in the end Intel gets paid for it. But I am happy as that will allow ATI GPUs to have thier own physics engine on the card.

As for AMD, they should have done a dual core first that could compete with Core2 and then focused on Phenom to enhance and make it better before release.

But Deneb and Shanghai should be their main focus in order to try to get that chips performanc at least to the lvl of Conroe/Penryn. I don't see any possible way to get it to compete with Nehalem without a completely new chip design or a major overhaul which they cannot afford to loose anymore time.

Then again I think the best way for AMD to compate with Intel is to reverse engineer Nelhalem and adapt it to socket AM3...

That should work as a stop-gap product until AMD can find a way to produce Shanghai...

What??

AMD need a wider issue cpu with more aggressive prefetchers and a much faster cache system.

Since Intel have better transistor arrays for their cache that isn't going to happen yet.

*opens door for technical engineer type to takeover discussion and wow the unwashed*

^ Not gonna happen

Jeez, if you don't know how to draw something (Not that the forum artist...), if you don't know how to make fast cache reverse engineer it (tell me, is that reasonably possible).

As you may know, the design and manufacturing of a cpu is a complicated process and takes years of planning in advance, talented engineers and lots of research and money. Now correct me if I'm wrong but AMD has a massive financial problem and therefore cannot afford to change their original plans and roadmap. It's no feasible to reverse engineer cache at this point.
Again correct me if I'm wrong, it took intel 3 years to release their comeback (although research and development had started before amd even released K8, a fact that many seem to forget), why would AMD, a much smaller company with a lot less money/engineers/fabs/...etc be able to come back faster?

This is all true really. Excpet Conroe existed much before that and Intel was just reworking it in a lot of ways. Heck the technology behind Conroe went back all the way to the Pentium Pro.

AMD themselves do not have the resources to turn things around faster than Intel though.

As for Nehalem to really say the truth is that Intel has been working on that sinc about the same time as Conroe was being worked on so it has taken them some time really egt to the point they are at now. I guess it could be given that Nehalem has been in the works for about 6 years or more even.

Well the bottom line is that if AMD is canceling future roadmaps and not releasing anything new that is worthy of really mentioning in the CPU department, this can be a bad signal to us consumers….expect to pay more for your CPU power in the future, As the gas crunch goes so will Intel. Without AMD competing we are left with one choice. I am neither an AMD nor Intel fan, but realize that without competition Intel will historically rape your wallets. This also can mean that just possibly, AMD will shift its focus and develop GPU’s and other technologies that they can compete in and make money. So…what we should be asking….”Where is AMD heading” and hope that they can continue to compete in the CPU market in the very near future. My new build will have both Intel (CPU) and AMD (GPU) living inside my case. It is a sign for the future IMHO.

actual benchmarks while running 64bit applications on 64bit vista

http://www.bit-tech.net/hardware/2 [...] _b3_cpus/9

Didn't read through the whole article but give and take between amd/intel chips performance wise.

To clarify, AMD came out with a NATIVE-quad core first, i never said at any time that intel didn't put their double-dual core in a single package short-cut "quad" out first.

AMD did however introduce 64bit cpus, dual-cores, IMCs first. No, intels failed attempt at an IMC between the p3 and P4 generations does not count no matter how cleverly they try to convince everyone they just decided their was no real benifit in having an IMC....but hey just because they haven't pulled an IMC in the 11 year gap since they first attempted it doesn't take away from the huge oppertunity they embraced after they scrapped years of work, RAMBUS...score!

I have never claimed that AMD had a lead in performance over intel. But it doesn't mean that just becasue intel has actually designed something worth buying 6 and a half years into this century that they are unequicically better than AMD in every regard. But i can see how my questioning the lack of benchmarks showing the performance of Intels 64bit chips in 64bit applications while AMD 64 bit benches are abundent makes me a fan boy. That's not remotely hypocritical.

Intel has been notoriously outspoken in denouncing the need for 64 bit OS's, applications and even hardware support. WHy? Because Intel doesn't play well at that code level. Which is why thier first 64bit chips were near identical to AMD, they had to licsence most of the 64bit extensions and egineering specs from AMD. I'm sure that was a tasty treat for intel to swallow, no reason they'd want to downplay 64 bit architecture after that.

When run on an AMD 64 bit single core 1.8ghz semperon Vista ..as in the OS as a whole is 11% faster running the 64bit version. Seems to me that when the OS, in which all application 32 or 64bit, is running 11% faster..you'd get a ripple effect. But how could that be relevant, right? I mean, it's not like system performance is what matters.

Oh wait,,,, we're talking about .... hmm ... what was it again ...oh yeah hardware! Now, what would performance have to do with different kinds of hardware...hmm...

Oh right. Everything.

I admit that i don't like intel as a company. I don't like their bussiness practice, i don't like their marketing strategy, i don't like the lack of design improvements that they present every 3 months with each new socket design, and complete platform.

I'm not impressed with the fact that while the core architecture may change very very little, if at all, and the CPU pin count and layout may be completly identical, the increase of cache and 100-200mhz speed incriment increases require you to buy a brand new system everytime they rename a core.

The fact that they can take a $350-400 cpu, double the cache, increase the speed by a few hundred mhz, and maybe unlock the multipliers, for an overall performance gain of 20% and feel justified upping the price tag to $1500-$1800. Though i admit it is entertaining seeing so many people bragging about their "uber" overclock on thier $1500+ cpu that decreases in price by $500 when intel pulls the same trick again 2 months later.

It's just dejavu...Intel puts out something quite good, markets the hell out of it, undercuts the competitions credibility however possible, then amps up production to constantly have almost the exact same thing in newer shinier packaging double or triple the cost and incompatible with it's virtually identical predacessor.

There is no comparison between $1500+ 45nm quad-cores and $200 90nm-65nm quad cores. They are not priced for the same demographic, they are not meant to compete. When AMD put out thier quads in the sub $200 range, the similarlly clocked Intel quads dropped from $340 to $230. So instead of AMD getting reconition for being able to put out a chip for less than 2/3 of intel in the same performance bracket, it becomes "durp de durp...AMD sucks cause intel done got a 64bit chip dat bein' 5-10% faster on my sparkly 32 bit OS for only 20% more"

Intel is great for 32bit Vista and XP, for people that only need 2 gigs of system memory and like to play nothing but games and brag about the big clock speed and synthetic benchmark scores they obtain in their $3000 system.

But personally i don't see the fun in shelling out several grand for hardware that is pushed near max spec and cranking out an extra 10-15% performance. Nor do i see the point in holding on to a software platform that is a decade old, when the next software generation has been around for 5 years.

As for my previously stated rig, yes 3.6ghz 939 toledo core, anyone that pays attention to overclocking threads would note that 3.4ghz - 3.5ghz while rare exists albiet with a water cooling set up. I happened to luck out with the LCCBE stepping which was almost imiedietly reserved for 185x2 toledo core opteron chips. $120 chip clocked to 3.6ghz on air, that is something to be proud of, which is why i posted screenies on extremeoverclocking.com when i first got the chip.

Where i also posted screenies of my 165 x2 toledo core opty clocked from 1.8ghz to 3.1ghz on air.

My 2400+ barton mobile clocked to 3.185ghz with 270 FSB and ram timings of 2-2-2-5. Along with half a dozen other mobile bartons 2200+ -2600+ versions clocked from 245FSB - 265FSB and 2.9ghz - 3.06ghz on air.

I upgrade platforms every 2 years or so. I upgrade my video card once a year. Gone from the 6800 nvidia, to 850xtx PE, to the 1950xtx to a 8800GTx which flaked out shortly after getting and replaced with the 2900xt.

Now i've snagged a Phenom 9850x4 BE for $150 off ebay, a DFI 790FX M2R mobo for $130, 2x2 gigs of Cas 5 OCZ 1066 ram. Took 45 minutes to clock the cpu to 3.1ghz. Just waiting on my pair of 4870xt to crossfire to arrive, which from what i've seen stomp the $700 G280,,,,for $100 less.

Not a matter of whether i can upgrade my hardware every 3 months, i don't want nor need to. Half the fun is pushing your Sub $1000 rig to $3000 performance levels...any moron can shell out a few grand for excessly powerful hardware to keep XP running smooth.

In fact, that seems to be Intels marketing strategy.

Actually 64bit CPU's have been around a very long time maybe 1st to x86-64. Actually Intel beat AMD to the dual core benchmark by a few months with the Pentium 4's not like it matters though. They had IMC's dating back to the 386. As for RAMBUS well it's RAMBUS what more can be said.

No because even yet we are just starting to see OS driver and hardware costs come down to allow for the smooth 64bit migration, come to think of it how many OEM machines made for tom dick and harry have 4gig's +? So what evidence do you have to support that Intel doesn't execute 64bit code well? Considering they are the leader in compilers, profilers, and support I would have to think their platform support and maturity would be better than say AMD's. The instruction set x86-64 has to be identical to allow for compatibility across all platforms, with regards that AMD themselves didn't follow their own x86-64 designs that Intel built to exact specifications; which caused XP 64 early beta's to not run on Intel machines. Side note though all extensions to the x86 platform are covered in the cross licensing agreement AMD and Intel have, as well compatibility is a must something you seem to be disregarding to prop up another poorly thought out point.

What?

3 months and a new socket I think you are thinking 989, AM2, AM2+, and soon to be AM3.

AMD does the same thing so what’s your point?

1500-1800 wow where you pulling those numbers from? But I will be honest what your saying sure sounds like AMD's Black Edition processors which you forget to mention were out 1st.

Again you are going to have to show this trend preferably in a bar graph.

So why do you buy AMD? Can you name 100 desktop related 64bit apps that have come out over the last 5 years?

Well you got to be mad that you have to buy a new AM2 or AM2+ board then.

Oh so you can show us a CPUZ validation for you magical over clock? But form what I am reading you seem to run loops of preferred content so I am sure your good with Photoshop to.

You know what it makes money maybe you should go out and buy a few more AMD chips so they don't go under.

Word, Playa.

It all depends on the benchmarks. The K8 and Core micro-architectures are pretty different and you can get a huge spread in performance if you look at certain individual benchmarks. For example, specInt will show that a Core 2 is several times faster than a K8 (due to extremely fast integer performance of the Core) while ScienceMark shows K8s eating Core 2s at the same clock speed for lunch (due to mostly x87 FPU code in the benchmark and K7s' and K8s' x87 FPUs being very strong).

I think the longest-lasting x86 architecture was the P6. It was introduced in 1995 with the Pentium Pro and ended in 2006 with the last Core Solo and Core Duo chips shipping. There were a few tweaks to P6 along the way, such as adding a few SIMD instructions (MMX to the PII, SSE to the PIII), changing the FSB (Pentium M, Core), and putting two of them on a single die (Core Duo) but they are all very similar internally.

A five-year lifespan for an x86 CPU architecture isn't all that odd if you look back at the past:

Intel
1978-1982: 8086 (4 years)
1982-1986: 80286 (4 years)
1986-1989: 80386 (4 years)
1989-1994: 80486 (4 years)
1993-1997: Pentium/P5 (4 years)
1995-2006: P6 (11 years)
2000-2006: NetBurst (6 years)
2001-2002: Itanium (1 year)
2002-present: Itanium 2 (6 years so far)
2006-present: Core (2 years so far)
2008-present: Atom (a month or two so far)

AMD
1995-1996: 5x86 (1 year, although AMD continued to sell 5x86s until 1999)
1996-1997: K5 (1 year)
1997-2000: K6 (3 years)
1999-present: K7 (9 years so far)- note that since 2005 the only K7s sold are embedded Geode NXes.
2002-present: Geode GX2 (6 years so far)
2003-present: K8 (5 years so far)
2007-present: K10 (a little over half a year so far)

Your proposed HTPC setup is not really an HTPC but a gaming computer that can play video and audio. An HTPC is supposed to be a machine that is designed primarily for playing back AV media using as low-powered, cool, and quiet components as practical. This means OCing is out unless you have an older system as your HTPC that *needs* to be OCed to yield acceptable performance. Many HTPC users use pretty aggressive power-management schemes to lower clock speed and voltage to control heat and noise to make the system as unobtrusive as possible when watching recordings or listening to music. An extremely powerful GPU is also not needed and is also just more heat and noise to contend with. A lower-end modern GPU typically has enough grunt to output a nice picture at 1920x1080p via XVideo or Direct3D and can do quite a bit of decoding and motion compensation offload from the CPU.

AMD buying ATi probably couldn't have come at a worse time for AMD's short-term finances but it is starting to pay off in much-improved products and platforms. It should end up being a good investment as ATi is doing pretty well right now and also having a chipset and GPU maker in-house lets AMD rival Intel in the lucrative business market.

The first 64-bit CPU ISA I can think of is that is not basically a one-off for supercomputers is MIPS64, which debuted with the R4000 in 1991. Intel's first 64-bit CPU was the Itanium in 2000, which uses the IA64 ISA.

I have a socket 939 X2 4200+ and a C2D U7500, and use a Q6600 at work. All three run x86_64 OSes and none of them have any trouble running their OSes nor letting any program address >4 GB memory.

There are more than 100 desktop-related 64-bit applications here and all are available for multiple 64-bit architectures:

http://packages.debian.org/stable/gnome/
http://packages.debian.org/stable/kde/

I've been running both of my machines on x86_64 since I first got them two and a half years ago and a year ago, respectively. With the exception of ETQW and Flash player, both run only x86_64 applications. The last time I had to run any other 32-bit application on either system because an x86_64 version was before OpenOffice.org got ported to x86_64 about two years ago.

That’s not what I was trying to say, I was saying that at the consumer level 64bit doesn't matter tom dick and harry don't need or seem to care if its 32bit or 64bit.

More so consumer 64bit application open source moves along but the money making market still is hesitant if not completely lazy on the transition to 64bit applications.

Word, Playa.