Wow, Price vs Perf right now in PC's is nice...

[Albuquerque]

Did you ever heart about buffering?

What about buffering? Buffered ram has no bearing on this conversation, so I have to assume you're talking about something else. And since no amoutn of software buffering is going to be affected by a piece of hardware which is running faster than the interface that allows it to communicate to the rest of the machine, then that too is a moot point.

So, please, tell me about buffering...

 

[Putas]

about that one in northbridge...

 

[Kyyla]

The Vista seems like a waste, why not XP Pro, or if you have an existing license why not use it? I won't touch vista for now, not even sure later. I am happy with linux and xp using virtualization for now(kernel virtualization modules).

Why would you use a similarly priced older OS on a new pc?

 

[Albuquerque]

about that one in northbridge...

What are you talking about? A one-sentence reply isn't sufficient to tell me.

 

[Putas]

It was enough to confirm my assumption
Most simply put, there is fifo between read and write ports of NB to cover this. How robust it is depends, but on modern chipsets you can be sure there is atleast so much to not loose performance when increasing memory clocks, although going asynchro. That is exactly the case bearlake exhibits (and step down from provious intels northbridges). As clocks are getting more diversed, peaks decreased and that is why performance gains after 400 MHz memory are increasing. GPU's have perhaps best implemented buffers, as there you rarely see any differrence between sync/async and even between various async ratios. Long pipeline helps a lot but anyway...

 

[3dilettante]

Buffering doesn't eliminate the latency penalty for synchronizing out of phase clock domains. If clock cycle is incomplete in the DRAM clock domain when the FSB clock is ready (that is, mismatched base clocks), it is not safe to pull data from any buffer that is receiving data because the DRAM bus may not have settled at the final state.

Whatever margin there is between the clock domains falls on the receiving side as a duty cycle (or more if there is a wide disparity), if it is faster, or a slight increase in overall latency because of the extra buffering.
It's a small penalty, but it is one that exists in most cases regardless.

Buffering also does not fake peak bandwidth. If the FSB is already saturated, the RAM's extra bandwidth has a limited effect if the northbridge services the occasional DMA request.

The problem with counting on extra bandwidth for that is that it is possible in FSB-limited situations that the CPU may be involved in orchestrating some of those DMA transfers or the CPU is working on critical data that determines what other memory clients should do. This means the slowed execution caused by the FSB causes DMA traffic to slow anyway.

Why else do most benchmarks show such limited gains to memory bandwidth that exceeds the FSB's bandwidth? Usually, the CPU is doing something pretty important when the FSB is saturated. It's only under some limited circumstances that other components can continue without rapidly running out of commands the CPU should be generating.

 

[Mize]

Why would you build a brand new rig, with high performance parts, and use a five-year old operating system on it?

Let's take a tally: we have people who don't understand the reality of power draw, who don't understand the reality and economics of "future proofing" a piece of PC hardware, and who can't really comprehend why a five year old OS isn't the perfect choice for a brand new PC?

Anyone want to chime in and ask me why I didn't spec out an ISA network card too?

You forgot the people with an over-inflated sense of self-righteousness and arrogance founded on a puffy chest on computer geek forums, but that's understandable since you wrote it Hard to see one's self clearly eh?

Forgive me almighty for putting trust in PSU calculators instead of running my rig through a frickin' ammeter! I am so shamed!

Oh and, while I'm at it, put me down as someone who thinks Vista is not the cat's meow. Ubuntu/SuSE and XPP on my rig. Stuck Vista on one guy's laptop at the office and it just crawls now. Shame.

 

[Albuquerque]

Call it what you like, but you're the one who was telling me the PSU was going to be short -- and used anecdotal evidence to support your claim. Empirical evidence was exactly my point, and was the further point of the whole aysnchronous DRAM speed commentary too.

Before assuming you know based on assumptions, make SURE you know based on hard evidence. Or is that too puffy-chested for you?

Back onto the DRAM speed topic, no amount of northbridge buffering makes running the memory bus faster than the system bus any more logical. The truth is that bandwidth to and from the northbridge, REGARLDESS of buffering is a set amount (1333). Hanging ANY peripheral device that can transmit or receive data faster than that is simply wasteful.

It would be like putting a PCI-E 2.0 card on a Pentium MMX 133mhz machine. What in god's green earth would you do with that kind of bandwidth on a CPU that can't talk that fast?

 

[Mize]

Call it what you like, but you're the one who was telling me the PSU was going to be short -- and used anecdotal evidence to support your claim. Empirical evidence was exactly my point, and was the further point of the whole aysnchronous DRAM speed commentary too.

Before assuming you know based on assumptions, make SURE you know based on hard evidence. Or is that too puffy-chested for you?

Actually, relying on PSU calculators is not "anecdotal."
Anecdotal would be "the two times I've pushed upward of 80% rated capacity I've blown the PSU within 6 months of 24/7 operation"

True, but hardly statistically significant as it's "anecdotal."

It was just a suggestion in an attempt to be helpful. I don't know or like you well enough to break out my ammeter and take measurements. Sorry.

 

[Mize]

Back onto the DRAM speed topic, no amount of northbridge buffering makes running the memory bus faster than the system bus any more logical. The truth is that bandwidth to and from the northbridge, REGARLDESS of buffering is a set amount (1333). Hanging ANY peripheral device that can transmit or receive data faster than that is simply wasteful.

So you want to future proof the OS (Vista) but not the DRAM?

 

[Albuquerque]

So you want to future proof the OS (Vista) but not the DRAM?

How is Vista "future proofing" anything? It's the current OS, WinXP is not. If you want to compare it that way, WinXP is the i875-chipset or OSes and Vista is the P35. 875's work great for a lot of people, and perform admirably for what they are. But it's old technology, and it IS going away regardless of you or anyone else's favor or bias.

As for your continued commentary on PSU's, please also see the other three people who had the exact same complaint to levy against your suggestion. You don't need to load up an ammeter, quite a few places already beat you to the punch. You are assuming your PSU's died because of load, and you're further assuming that you were continuously loading them past 80%. I call that a complete logical fallacy on both accounts.

I've gone through a few power supplies in my life, and not all of them were PC power supplies. Here's a nifty fact about about electronic equipment that contains very large capacitors and coils: they degrade, and not just from overload. Most of what makes them degrade in real life isn't overcurrent, it's all the noise, ripples and voltage sags in your home's mains. Overload is a great way to make a PSU fail in a catastrophic way, but "life" is the more often measure of destruction.

Didn't I see you post in a nother thread that your 5v rail was sagging, and you had to make some adjustments to get it working again? Were you overloading your 5v rail with -- uh -- what the hell uses 5v rail again? I bet you weren't, because almost nothing uses that rail anymore. it's wear and tear, and that's what happens to electrolytic capacitors.

 

[Mize]

Hey ABQ, this is way off thread. If we really want to go hog wild I can derive dielectric breakdown strength based on polymer and geometry. I'm a materials scientist specializing in sensors. My wife's a chemist who did embedded capacitors for years. The 5v rail was sagging because it was shared. Adjusting the 12v shouldn't affect the 5v in a good supply unless something's wrong. Whatever. I just don't skimp on PSUs as I've lost other components as a result. And the idea that I'm wrong for not scanning the hardware sites before suggesting you consider a beefier one is wild. I've always viewed B3D as a conversational place and only made a suggestion. Next time I'll include some disclaimer about not conducting a full research project before posting...k?

As to Vista...bah...it's your preference. You can go to Office 2007 and even more new proprietary formats for your docs and mailboxes too. Your choice. I'm tired of being locked-in by Microsoft and their kludgy software (an Outlook search took me several minutes compared to 10 seconds using mbox format under evolution). My preference.

 

[Blazkowicz]

overspeccing the ram a bit doesn't hurt, even to still run it at 667. it buys you some tranquillity. I suffered from extremely annoying ram issues, had to junk my 768MB sdram, got a new mobo and 512 DDR, later had to revert to 256MB with underclock to keep stability, aargh!
same reason you overspec the PSU.. which is reasonable done with that 450W. and Fortron ones are especially good for their displayed wattage, I take a 350W Fotron over a cheapo 550W any day.

as for the OS, who cares. when you've got something modern your experience is more determined by the settings and what apps you use (no matter the OS I'll use the same firefox, gaim, open office.org etc.). XP is quite decent, supporting like 128GB ram and 32 CPU when it's called server 2003 x64. what it lacks is the new I/O prioritization, directX 10 and the superfetch, indexing stuff etc. but apart from DX10 gaming, why should I care if everything was already fast anyway with XP on a 5 year-old computer. get either OS.

 

[Mize]

I like this "normalizer" from the eXtremeTech PSU calculator:

4 Electrolytic capacitor aging. When used heavily or over an extended period of time (1+ years) a power supply will slowly lose some of its initial wattage capacity. We recommend you add 10-20% if you plan to keep your PSU for more than 1 year, or 20-30% for 24/7 usage and 1+ years.

Clearly just a fudge factor, but it basically shows that PSUs slowly die from day 1 so overspec'ing the PSU may simply buy you longevity, eh? So what if I have 24/7 heavy use for 3, 4 or 5 years?

 

[Albuquerque]

Hey ABQ, this is way off thread. If we really want to go hog wild I can derive dielectric breakdown strength based on polymer and geometry. I'm a materials scientist specializing in sensors. My wife's a chemist who did embedded capacitors for years. The 5v rail was sagging because it was shared. Adjusting the 12v shouldn't affect the 5v in a good supply unless something's wrong. Whatever. I just don't skimp on PSUs as I've lost other components as a result. And the idea that I'm wrong for not scanning the hardware sites before suggesting you consider a beefier one is wild. I've always viewed B3D as a conversational place and only made a suggestion. Next time I'll include some disclaimer about not conducting a full research project before posting...k?

Welp, then someone in your claimed profession should already know the nonsense you were speaking. As such, it seems you have indeed at least realized this, so we'll leave it just like it is.

As to Vista...bah...it's your preference. You can go to Office 2007 and even more new proprietary formats for your docs and mailboxes too. Your choice. I'm tired of being locked-in by Microsoft and their kludgy software (an Outlook search took me several minutes compared to 10 seconds using mbox format under evolution). My preference.

Funny you mention Office 2007 in the same breath as "proprietary", since Office 2007 now uses an entirely open-source XML format as the default for essentially every document type it creates. Remember my sentence earlier about assuming versus hard evidence? Yeah, you might watch your toes...

 

[L233]

Maybe, but then why limit yourself to DDR2-800 when DDR2-1066 is just that tiny bit much more? Here's a reason why not: because DRAM prices are going nowhere but down. Why buy something today that you're NOT going to use due to the notion of "future proof"; versus buying an equivalent or better product when you're actually going to use it and it will actually be cheaper?

Because it's unlikely that there will be anything with a FSB > 1600 that will still be compatible to the mainboard. With DDR2-800, you're on the safe side for a paltry amount of money. And buying DDR2-800 now is still cheaper than buying DD2-667 now plus DDR2-800 later, even if prices continue to drop.

I do agree with your general sentiment that buying "future proof" is often a misguided idea but in this case, the added cost is so low that I'd rather err on the safe side.

As for 3Ghz Penryns dual-cores going for $180 in January? I'll believe it when I see it.

20th January, from what I've read.

 

[Mize]

Welp, then someone in your claimed profession should already know the nonsense you were speaking. As such, it seems you have indeed at least realized this, so we'll leave it just like it is.

Oh you mean that PSU's degrade from day one and that over-spec'ing means longer life? Yeah, check, got it.

 

[Albuquerque]

Oh you mean that PSU's degrade from day one and that over-spec'ing means longer life? Yeah, check, got it.

So then, now that you've asserted that, and we've already determined that this system soaks up somewhere around 330W under 100% load, and I've specced a high-quality (from ALL accounts) power supply rated for 450W -- then I've overspecced by some 33%.

Care to continue arguing about your nonsensical stance on this subject?

Because it's unlikely that there will be anything with a FSB > 1600 that will still be compatible to the mainboard. With DDR2-800, you're on the safe side for a paltry amount of money. And buying DDR2-800 now is still cheaper than buying DD2-667 now plus DDR2-800 later, even if prices continue to drop.

My point was this: this isn't MY machine, this machine belongs to my boss. His upgrade schedule is something along the lines of "when this one stops working". His old machine was a P4 / 1.4 and 512mb of ram. The 40gb drive finally puked a few months ago, and he's finally decided to buy a new one.

This isn't some enthusiast that's going to replace CPu's every six months. So when THIS computer dies in three or four years, whill he replace the CPU with a 400mhz bus part? I'd wager not. And even if he did, would LGA-775 even still be current? I'm quite sure not.

Even if he did upgrade processors and video cards n 18 months, will he notice the speed increase from DDR-667 to DDR-800 in anything he does? The new processor (probably 2x faster) and new video card (also probably 2x faster) are going to FAR outweigh the extra 133mhz of ram speed.

Now, sometime next year when I build my own rig? Sure. I'll be buying some memory that's quite a bit faster than 667; hell I'll probably go for at least 1066 if not more. But I plan on overclocking my rig for all it's worth, and I won't know the ceiling until I get there. In that case, overspeccing is a good idea.

The uses that my 50+ year old boss will find for this machine do not necessitate or even suggest the need for overspecced memory.

 

[Mize]

Funny you mention Office 2007 in the same breath as "proprietary", since Office 2007 now uses an entirely open-source XML format as the default for essentially every document type it creates. Remember my sentence earlier about assuming versus hard evidence? Yeah, you might watch your toes...

Absolutely false. Microsoft's new "open" XML format cannot be read by the last version of Office without an add-on and support outside MS products is sparse at best (less than beta).

Their Outlook database is not XML or open in any way. MS's XML is not approved (failed the standards ballot).

And tell me, when you take that graphic-intensive MS Word doc in "open" XML over to 2003 or O how does it look

 

[Mize]

So then, now that you've asserted that, and we've already determined that this system soaks up somewhere around 330W under 100% load, and I've specced a high-quality (from ALL accounts) power supply rated for 450W -- then I've overspecced by some 33%.

Care to continue arguing about your nonsensical stance on this subject?

I'm not arguing about your PSU choice - I'm arguing about you acting like an asshole when someone makes a friendly conversational suggestion based on an unwillingness to do extensive research on your behalf. Generally a "I researched it and I'm confident I have a good enough PSU" goes over better than thumping your chest and insulting those kind (or in this case dumb) enough to try to help a jackass.