If this is your first visit, be sure to check out the FAQ by clicking the link above. You may have to register before you can post: click the register link above to proceed. To start viewing messages, select the forum that you want to visit from the selection below. |
|
|
Thread Tools | Display Modes |
#21
|
|||
|
|||
On Fri, 15 Oct 2004 23:46:38 GMT, gaffo wrote:
JK wrote: There are those with 16 bit processors that still perform flawlessly, although some people want to run the latest and highest performing software, which in less than a year will probably be 64 bit software for most applications. HAHAHAHAHAHAHAHAHAHAHAHAHAHA!!!!!!!!! Talk about Beachfront property! How long did it take 16-bit to yield to 32-bit?.................... no, don't bother, I'll tell you, effectively TEN YEARS!!!!!!! 386-1986............................32 bit software showed up in 1996. 64-bitness will remain irrelivant WRT to the home user for another 8 yrs or so. (using the Opteron initial release time). and yes - I beleive this is a realistic timeframe for "Joe Ave" (i.e. the mainstream). I really don't think this is an accurate time-frame this time around. This are already very different. It took 6 years for the very first 32-bit x86 operating system to be released after the 386 came to market, this time around we had the first 64-bit OS for the Opteron only 6 months after it's release. Win95 brought the first 32-bit (err, kinda) OS to the mainstream desktop 9 years after the release of the 386, now we're probably looking at about 2 years after the Opteron release for WinXP x64 to come out. On the application site, there are already TONS of open-source programs ported to 64-bit in Linux-land, and even in Windows world there are a handful that are available now (even before the OS is available). In short, the transition is happening MUCH faster now than it did back in the 386 days. Just as in the late 80's and early 90's (i.e. 16bit on a 32bit) million of us will be using a 64-bit chip to run 32 bit software. Sure, especially in Windows. We really don't need a 64-bit version of most applications, though the performance benefits (in the specific case of x64) might see them arise. However there are enough applications that WILL see 64-bit become the standard reasonably quickly that I think it's rather short-sighted to buy a 64-bit system if building one today. Given that you can get a 64-bit processor for "free" (ie same cost as a comparable 32-bit chip), then the question isn't so much "why 64 bits?" as "why not 64 bits?" Now a 64-bit chip means nothing to me, and on board memory controller which will let me run 32-bit/16-bit faster DOES MEAN SOMETHING.......in the "here and now". I'll buy for 64-bitness in the next decade - thanks. I'll buy 64-bit when I can afford to buy a major upgrade, because honestly there's no good reason not to these days. ------------- Tony Hill hilla underscore 20 at yahoo dot ca |
#22
|
|||
|
|||
Johannes H Andersen wrote:
JK wrote: [...] The move to 64 bits and memory controllers integrated into the cpu for much greater performance are important reasons to upgrade. Of course those chips with integrated memory controllers are made by AMD though. How do you conclude that integrated memory controllers results in much greater performance???????? They (integrated memory controllers) lower the latency to main memory significantly. ergo the chip isn't spinning it's wheels aimlessly anywhere near as much in the event of an L2 cache miss. -JB |
#23
|
|||
|
|||
The move to 64 bits and memory controllers integrated into the
cpu for much greater performance are important reasons to upgrade. When will we see this? What impact does the cancellation of the 4 GHz processor have on the 925XE, if any? It'd be nice if they had a timetable. |
#24
|
|||
|
|||
On Fri, 15 Oct 2004 20:53:06 GMT, Johannes H Andersen
wrote: JK wrote: [...] The move to 64 bits and memory controllers integrated into the cpu for much greater performance are important reasons to upgrade. Of course those chips with integrated memory controllers are made by AMD though. How do you conclude that integrated memory controllers results in much greater performance???????? sighBecause latency is the enemy... in every part of a system - losing the North Bridge FSB connection between CPU and memory controller counts for a lot. The fact that the North Bridge memory address arbitration and routing is moved into the CPU die and done at CPU clock speeds is another big factor. And why is it important to upgrade when my Dual Channel Hyper Threading Northwood 2.8/800 has performed fast flawlessly since I build it in July?????? Idle Temp 36 Deg. Max temp with 100% flat out numerical analysis = 56 Deg ???????? If you're happy with what you have thats err, fine. That Hyperthreading is there, in large part, of course, to counter memory latency. Rgds, George Macdonald "Just because they're paranoid doesn't mean you're not psychotic" - Who, me?? |
#25
|
|||
|
|||
On Fri, 15 Oct 2004 23:46:38 GMT, gaffo wrote:
JK wrote: There are those with 16 bit processors that still perform flawlessly, although some people want to run the latest and highest performing software, which in less than a year will probably be 64 bit software for most applications. HAHAHAHAHAHAHAHAHAHAHAHAHAHA!!!!!!!!! Talk about Beachfront property! How long did it take 16-bit to yield to 32-bit?.................... no, don't bother, I'll tell you, effectively TEN YEARS!!!!!!! In your little cell, maybe! 386-1986............................32 bit software showed up in 1996. Uhh that was Windows 95 - there had been 32-bit software available, even from M$, well before that and of course M$ did have the little detour known as OS/2 which was 16-bit in its original form. Before the 32-bit GUIs were available, people were running 32-bit software routinely - some of them without even knowing it - in the form of DOS Extenders from Phar Lap et.al... and often (pre-emptively) multi-tasked under the DesqView environment. That was in 1987/8 for Phar Lap - I know... I used it. 64-bitness will remain irrelivant WRT to the home user for another 8 yrs or so. (using the Opteron initial release time). Obviously you have not grasped what 64-bitness means. and yes - I beleive this is a realistic timeframe for "Joe Ave" (i.e. the mainstream). Just as in the late 80's and early 90's (i.e. 16bit on a 32bit) million of us will be using a 64-bit chip to run 32 bit software. Now a 64-bit chip means nothing to me, and on board memory controller which will let me run 32-bit/16-bit faster DOES MEAN SOMETHING.......in the "here and now". I'll buy for 64-bitness in the next decade - thanks. Cut your nose off to spite your face if you want. snip of boilerplate Rgds, George Macdonald "Just because they're paranoid doesn't mean you're not psychotic" - Who, me?? |
#26
|
|||
|
|||
gaffo wrote:
How long did it take 16-bit to yield to 32-bit?.................... no, don't bother, I'll tell you, effectively TEN YEARS!!!!!!! 386-1986............................32 bit software showed up in 1996. The 386's featureset was already being exploited within a year of introduction. Remember all of those DOS extenders and memory managers were also 32-bit software. They added a 32-bit "hypervisor" layer below DOS, long before Windows did the same thing. Qemm386 and Desqview were out within a year of the first 386. Then in Unix land, SCO (when it was still a real company) had Xenix and then eventually Unix running on 386s. Yousuf Khan |
#27
|
|||
|
|||
Johannes H Andersen wrote:
How do you conclude that integrated memory controllers results in much greater performance???????? And why is it important to upgrade when my Dual Channel Hyper Threading Northwood 2.8/800 has performed fast flawlessly since I build it in July?????? Idle Temp 36 Deg. Max temp with 100% flat out numerical analysis = 56 Deg ???????? Because it takes a P4 with about 2MB of L2 to equal the performance of an A64 with only 512K of L2. Yousuf Khan |
#28
|
|||
|
|||
Caecina wrote:
The move to 64 bits and memory controllers integrated into the cpu for much greater performance are important reasons to upgrade. When will we see this? What impact does the cancellation of the 4 GHz processor have on the 925XE, if any? It'd be nice if they had a timetable. Looks like Intel is having trouble producing P4's with the 1066Mhz FSB, even though the 925XE chipset is ready to go. BTW, 925XE is just a more thoroughly tested 925X chipset. If you're going to buy a P4, then buy with the 925X right now, don't bother waiting around for the 1066Mhz FSB devices. Or just go with the alternatives such as AMD -- they've been hitting their targets more or less most of the time for the past several years. PCI Express chipsets are available for them now as well, from ATI, nVidia, VIA, and SIS. Yousuf Khan |
#29
|
|||
|
|||
Or just go with the alternatives
such as AMD -- they've been hitting their targets more or less most of the time for the past several years. PCI Express chipsets are available for them now as well, from ATI, nVidia, VIA, and SIS. I'm definitely considering AMD more than I would've, had Intel worked out it's problems with the Pentium 4. However, the 512k cache bugs me. It seems outdated. I mean, I've had a comps. w/ that amount of cache for about five years. |
#30
|
|||
|
|||
Caecina wrote: Or just go with the alternatives such as AMD -- they've been hitting their targets more or less most of the time for the past several years. PCI Express chipsets are available for them now as well, from ATI, nVidia, VIA, and SIS. I'm definitely considering AMD more than I would've, had Intel worked out it's problems with the Pentium 4. However, the 512k cache bugs me. It seems outdated. Not really. The Athlon 64 Clawhammer, which was the original Athlon 64 had a 1 meg L2 cache and single channel memory access. The next series of Athlon 64 was the Newcastle, which moved to 512 K of L2 cache, but still has single channel memory access. If one compares a Clawhammer 3200+ to a Newcastle 3200+, the performance is almost identical, however the Clawhammer ran at 2 ghz, while the Newcastle runs at 2.2ghz. So the extra 512K cache in the Clawhammer only equalled around a one speed grade increase in terms of clock speed. The latest Athlon 64, the Winchester, has dual channel memory access. The Winchester 3200+ runs at 2 ghz. The extra memory controller adds around the same performance as increasing the clock speed by one speed grade. There are currently Athlon 64 chips being made with 1 meg of L2 cache, the Athlon 64 4000+, and the Athlon 64 FX-55. Those are around $800 and $920 though. I mean, I've had a comps. w/ that amount of cache for about five years. |
Thread Tools | |
Display Modes | |
|
|
Similar Threads | ||||
Thread | Thread Starter | Forum | Replies | Last Post |
4.0Ghz P4 now officially cancelled | Yousuf Khan | General | 252 | November 5th 04 07:22 PM |
4.0Ghz P4 delayed | Yousuf Khan | General | 1 | July 31st 04 12:47 AM |
4.0Ghz P4 delayed | Yousuf Khan | Intel | 1 | July 31st 04 12:47 AM |
Partition Hex Codes? | NiteWolf1138 | General | 2 | February 11th 04 05:53 PM |
Canon 'officially resealed' printers | Si | Printers | 4 | September 2nd 03 12:32 AM |