Folding is a program that runs forever in the background of your PC, taking up less than 10MB of RAM, and using your processor only when it's available. It manipulates synthetic representations of proteins found in the body to determine the cause of diseases like alzheimer's and cancer.
When certain bodily proteins "Fold" wrong, people get sick. We're seeing exactly HOW they fold wrong, so we can reverse the process.
Your Front Side Bus is what your SYSTEM runs from, like you AGP Card runs 66MHz for the AGP Transfer speed, yet the GPU runs at a much higher clock speed. Your CPU executes information while sending out information. So basically if you have a Multiplyer of 18 and a FSB of 133, then your CPU is working at a frenquency of 18:1 or 2.4GHz. But your system feeds off of only the FSB speed which is then broken down, altered, you name it.
How the heck do you convince an intel fanboy that he can stick his bandwith where the sun don't shine? (I do not intend to do it physically )
He is going bananas bacause of his high bandwith.
And futhermore he is joking about "old AMD's" and their abysmal bandwith"
He is a moron.
But do you guys have any words of wisdom I can use?
He has an "old" P4 NW
First off, what type of memory does he have? If it's SDR or even RAMBUS: make fun of what a huge bottleneck his memory is to his system bandwidth. "The bus is only as fast as the slowest contender" -- very true.
#1) Show him how bandwidth is measured.
• MHz of Bus (not Double or Quad pumped) * Bus Width (in bytes) * 2 (if DDR applies) * # of channels (if applies)** = MB/sec
• Do that for Memory & CPU of his and your system. That's the MAXIMUM theoretical limit of each piece of hardware.
** Multi channel memory performance depends also on FSB. This complicates things a bit.
#2) Theoretical.. is theoretical.
• It's a known fact that the Athlons are more efficient than P4s. The figure varies as each chipset and setup in its entirety is different. We want some numbers in the field so: Athon XP - 92% vs. P4 - 84%.
• Take the maximum theoretical bandwidth * efficiency %. Now, that's just the maximum said system will hit, not the average.
#3) Then, you poke fun at the fact you can lower your multi to get more FSB if you wanted (and hardware permitting). He will have to just cross his fingers on FSB OCing.
#4) Dig more into memory. DDR is hands down the type to have. SDR is... slow! And RAMBUS isn't so hot. It uses a very small BUS width, ramped MHz, high latency, and dual channel to even survive early in the game. No headroom anymore... :woowoo:
IF THAT DOESN'T WORK
Just run Sandra benchies to give him a visual.
Well he has DDR ram, but I will look into the bandwith issue.
He is referring to a SiSoft Sandra test, showing his theoretic double bandwith.
A DC P4 system does indeed have a higher bandwith score in Sandra.
In my case the equation would look something like this:
225*128*2*2 is that correct?
EDIT: It has to be: 225*8*2*2 instead.
Result 7200
BTW I found this great explanation concerning memory and cache.
This seems like a good example, but it has a serious flaw.
In the example, the 10 cars on the street are indeed the ram, and doubling the data rate is like increasing the entrances into the lot from 1 to 2. That's what the dual channel memory bus does, and it does stand to reason that the bandwidth is now twice what it was before.
However the CPU runs 'most of the time' out of the internal cache memory, which is very high speed memory indeed. You typically have 256Mb or 512Mb of main memory, but the cache memory is much faster and much smaller and lately its built in the chip with the cpu itself, like 64kb to 512kb depending on the CPU. The cars are the main memory, the parking lot is the cache memory. The CPU 'internal' bus to the cache memory is not mentioned in the example, and that is like having 10 entrances from the other side of the parking lot going to the CPU. The CPU never goes out to the street to get its data, it gets it from the parking lot through its 10 entrances at very high speed. Only on occasion does it find that what it needs isn't in cash (the parking lot with 10 entrances) and has to go out to main memory to read in a few more blocks of ram (the street with 2 entrances).
So the CPU spends most of its time running from the cache memory from a very high speed bus, and only on occasion does the cache memory need to be updated with main memory.
So it makes complete sense why doubling the bandwidth to main memory only increases overall performance by 5% or so, because it's only accessed 10% of the time, and doubling it's bandwidth then cuts that bottleneck time in half, saving you half of the 10% of time spent doing that operation, which is 5% of the system performance.
Get an athlon 64! this one i got today is up and running wtih no hicups and is FAST!!!!! I have yet to benchmark in more than 1 game but in UT my FPS hit 1700 and average stayed around 700
I know but I am defending the good old Thoroughbred and Barton.
I wouldn't buy an Athlon 64 any time soon though.
Maybe late 2004/early 2005 it isn't mature yet.
And there isn't much room for any OC.
Well he has DDR ram, but I will look into the bandwith issue.
He is referring to a SiSoft Sandra test, showing his theoretic double bandwith.
A DC P4 system does indeed have a higher bandwith score in Sandra.
In my case the equation would look something like this:
225*128*2*2 is that correct?
EDIT: It has to be: 225*8*2*2 instead.
Result 7200
His results are higher than yours? His results are higher than a referenced system in Sandra? Your system is slower than a P4 in Sandra? Which is it?
Make sure you compare his to yours, don't reference if you're trying to pwn him.
In your edit, you got the bandwidth right. That's how you do it.
Does he have DualChannel DDR? And what speed of DDR and CPU FSB?
What about your system? Same questions...
Let me know, and I could make a spiffy chart with every mathematical equation figured.. and a winner highlighted... if you want
The only thing the P4 has going for it is the Quad-Pumped FSB. Along with that (or just normal Intel engineering ), lower efficiency than Athlons is present. Thus, a repeat from Intel: A heavy aspect on hardware not always the best.
Well I have the system mentioned in my sig.
Oops didn't have FSB mentioned.
225 MHz FSB sync at 2.0-2-2-11 multiplier 9.5
He has a P4 NW 2.6 800 MHz pumped FSB(real fsb 200)
Dual channel PC3500 at 200 MHz.
As far as I know he isn't running his memory at PC3500 speed.
Doesn't matter though, my PC3200 is still faster at 225 MHz.
I am not referencing, he is, I tried an "onboard assault".
Mentioning that DC is good for onboard sound and GFX. A desperate measure, I know.
And yes I am trying to pwn him, he has annoyed me a bit too much.
What is even worse, is that he is imagining things.
He says his present P4 sys is faster than the Barton at 2.5 Ghz he owned before using a NF-7.
The guy is seriously sick.
I just want to add a few things concerning the first post (or the topic). I have had a few Athlon XP systems (1600 - 1800 - 2100 - 2x2500's) and currently own one (2400m). I also have (See sig) a P4 system. Each has its own advantages. One thing I have noticed is that the AMD loads up windows faster (not a WHOLE lot, but noticeable) on identical instals (cept drivers or course :P). At first I thought hyperthreading was just another marketing ploy, but it does actually work (or so it seems), it feels alot smoother opening/running multiple programs on the P4 than on the Athlon. For gaming, well after a certain point, its hard to tell the difference. The Athlon seems to load maps and levels or whatnot, a little bit faster (just a little). Ripping/encoding music and dvd's the P4 is faster (I don't know what benchmarks show, but personal experince show the P4 is faster). The Athlons run hotter. Some of you may argue this, but I saw it with my own eyes. There was something else .... oh well I forgot.
To sum things up: With the speed of the top end athlons and p4s, for gaming you can't really tell. For office programs and such, its trivial. And for media, encoding especially, the P4 is a little faster. Overall I would recommend the Athlon just because its so much cheaper. If they were priced the same it would be a tough desicion.
His results are higher than yours? His results are higher than a referenced system in Sandra? Your system is slower than a P4 in Sandra? Which is it?
Make sure you compare his to yours, don't reference if you're trying to pwn him.
In your edit, you got the bandwidth right. That's how you do it.
Does he have DualChannel DDR? And what speed of DDR and CPU FSB?
What about your system? Same questions...
Let me know, and I could make a spiffy chart with every mathematical equation figured.. and a winner highlighted... if you want
The only thing the P4 has going for it is the Quad-Pumped FSB. Along with that (or just normal Intel engineering ), lower efficiency than Athlons is present. Thus, a repeat from Intel: A heavy aspect on hardware not always the best.
I would indeed like such a chart, if you are willing to make one.
Thanks in advance.
I just want to add a few things concerning the first post (or the topic). I have had a few Athlon XP systems (1600 - 1800 - 2100 - 2x2500's) and currently own one (2400m). I also have (See sig) a P4 system. Each has its own advantages. One thing I have noticed is that the AMD loads up windows faster (not a WHOLE lot, but noticeable) on identical instals (cept drivers or course :P). At first I thought hyperthreading was just another marketing ploy, but it does actually work (or so it seems), it feels alot smoother opening/running multiple programs on the P4 than on the Athlon. For gaming, well after a certain point, its hard to tell the difference. The Athlon seems to load maps and levels or whatnot, a little bit faster (just a little). Ripping/encoding music and dvd's the P4 is faster (I don't know what benchmarks show, but personal experince show the P4 is faster). The Athlons run hotter. Some of you may argue this, but I saw it with my own eyes. There was something else .... oh well I forgot.
To sum things up: With the speed of the top end athlons and p4s, for gaming you can't really tell. For office programs and such, its trivial. And for media, encoding especially, the P4 is a little faster. Overall I would recommend the Athlon just because its so much cheaper. If they were priced the same it would be a tough desicion.
Agreed, I defy any one to tell the difference between the two in a blind performance test running the same equal speeds and applications especially if they are above 2 ghz CPU's.
It's more a money and entheusists thing than a real world usage issue... IMHO.
"g"
Like croc, I also have both XP and P4 systems and really and truly, there isn't much difference between them as far as speed, even though the actual GHz numbers are much higher for the P4. I do find that the P4 is a little smoother when multitasking, probably due to hyperthreading. The P4 I have will also fold Gromacs work faster than a single proc AMD system, also due to hyperthreading, but the AMD will run rings around the P4 when folding Tinker work because Tinker WU's don't use SSE, which the P4 must use for FP work to progress quickly. The AMD system is cheaper to put together and it is also easier to overclock, with an unlocked proc. To overclock a 2.4 or 2.6 P4, you will need very high end memory to deal with the ultra fast fsb speeds needed for the high overclocks. The P4's memory bandwidth is much higher in Sandra than an XP, but I don't play Sandra, just games where I really don't see a big difference in either.
The P4's memory bandwidth is much higher in Sandra than an XP, but I don't play Sandra, just games where I really don't see a big difference in either.
LOL I don't play Sandra, that's a good one moddocktor.
Comments
Folding is a program that runs forever in the background of your PC, taking up less than 10MB of RAM, and using your processor only when it's available. It manipulates synthetic representations of proteins found in the body to determine the cause of diseases like alzheimer's and cancer.
When certain bodily proteins "Fold" wrong, people get sick. We're seeing exactly HOW they fold wrong, so we can reverse the process.
for example
athlon 1700+ runs at
133fsbx11 multiplier which makes it 1463 mhz
and if u change the multiplier to 12
u will get 1596 mhz.
I know the FSB speed is the actual speed of the FSB, so what does the multipier measure?
Oh ok. So it's just a measurement.. That works
Think of it as the CPU:FSB ratio. In my case, 12:1.
Your Front Side Bus is what your SYSTEM runs from, like you AGP Card runs 66MHz for the AGP Transfer speed, yet the GPU runs at a much higher clock speed. Your CPU executes information while sending out information. So basically if you have a Multiplyer of 18 and a FSB of 133, then your CPU is working at a frenquency of 18:1 or 2.4GHz. But your system feeds off of only the FSB speed which is then broken down, altered, you name it.
How the heck do you convince an intel fanboy that he can stick his bandwith where the sun don't shine? (I do not intend to do it physically
He is going bananas bacause of his high bandwith.
And futhermore he is joking about "old AMD's" and their abysmal bandwith"
He is a moron.
But do you guys have any words of wisdom I can use?
He has an "old" P4 NW
#1) Show him how bandwidth is measured.
• MHz of Bus (not Double or Quad pumped) * Bus Width (in bytes) * 2 (if DDR applies) * # of channels (if applies)** = MB/sec
• Do that for Memory & CPU of his and your system. That's the MAXIMUM theoretical limit of each piece of hardware.
** Multi channel memory performance depends also on FSB. This complicates things a bit.
#2) Theoretical.. is theoretical.
• It's a known fact that the Athlons are more efficient than P4s. The figure varies as each chipset and setup in its entirety is different. We want some numbers in the field so: Athon XP - 92% vs. P4 - 84%.
• Take the maximum theoretical bandwidth * efficiency %. Now, that's just the maximum said system will hit, not the average.
#3) Then, you poke fun at the fact you can lower your multi to get more FSB if you wanted (and hardware permitting). He will have to just cross his fingers on FSB OCing.
#4) Dig more into memory. DDR is hands down the type to have. SDR is... slow! And RAMBUS isn't so hot. It uses a very small BUS width, ramped MHz, high latency, and dual channel to even survive early in the game. No headroom anymore... :woowoo:
IF THAT DOESN'T WORK
Just run Sandra benchies to give him a visual.
He is referring to a SiSoft Sandra test, showing his theoretic double bandwith.
A DC P4 system does indeed have a higher bandwith score in Sandra.
In my case the equation would look something like this:
225*128*2*2 is that correct?
EDIT: It has to be: 225*8*2*2 instead.
Result 7200
I wouldn't buy an Athlon 64 any time soon though.
Maybe late 2004/early 2005 it isn't mature yet.
And there isn't much room for any OC.
His results are higher than yours? His results are higher than a referenced system in Sandra? Your system is slower than a P4 in Sandra? Which is it?
Make sure you compare his to yours, don't reference if you're trying to pwn him.
In your edit, you got the bandwidth right. That's how you do it.
Does he have DualChannel DDR? And what speed of DDR and CPU FSB?
What about your system? Same questions...
Let me know, and I could make a spiffy chart with every mathematical equation figured.. and a winner highlighted... if you want
The only thing the P4 has going for it is the Quad-Pumped FSB. Along with that (or just normal Intel engineering
Oops didn't have FSB mentioned.
225 MHz FSB sync at 2.0-2-2-11 multiplier 9.5
He has a P4 NW 2.6 800 MHz pumped FSB(real fsb 200)
Dual channel PC3500 at 200 MHz.
As far as I know he isn't running his memory at PC3500 speed.
Doesn't matter though, my PC3200 is still faster at 225 MHz.
I am not referencing, he is, I tried an "onboard assault".
Mentioning that DC is good for onboard sound and GFX. A desperate measure, I know.
I am planning a Thrax assault. using Thrax's words of wisdom from this very thread.
Mentioning Prescotts "improvements" as well.
http://www.short-media.com/forum/showpost.php?p=50041&postcount=2
And yes I am trying to pwn him, he has annoyed me a bit too much.
What is even worse, is that he is imagining things.
He says his present P4 sys is faster than the Barton at 2.5 Ghz he owned before using a NF-7.
The guy is seriously sick.
To sum things up: With the speed of the top end athlons and p4s, for gaming you can't really tell. For office programs and such, its trivial. And for media, encoding especially, the P4 is a little faster. Overall I would recommend the Athlon just because its so much cheaper. If they were priced the same it would be a tough desicion.
I would indeed like such a chart, if you are willing to make one.
Thanks in advance.
Agreed, I defy any one to tell the difference between the two in a blind performance test running the same equal speeds and applications especially if they are above 2 ghz CPU's.
It's more a money and entheusists thing than a real world usage issue... IMHO.
"g"
LOL I don't play Sandra, that's a good one moddocktor.