ComputersFollow

Holy ****. If you sell anything at all at those ludicrously high prices, I'm totally in the wrong business. Your marking processors up 50% over retail and hoping no one notices?

The web site is functionally fine. The prices are absurd.

about the only thing i found reasonable were the prices for memory. everything else can be bought cheaper from a tiger direct retail store, and cheaper still on www.pricewatch.com web site.

i like the boxes, they look sweet. wouldnt pay more than 100 bucks for them though.

and WTF? no AMD? serious gamers know AMD whips intels **** when it comes to gaming benchmarks. its the computer geeks, not the gamers who buy intell for the latest and greatest and dont mind spending a few weeks upgrading drivers and fixing bugs. and incase you didnt research it, its the gaming market driving computer sales.

gona loose a good 40 percent of your potential market not selling AMD and Radeon. and also if you didint know yet, they are one and the same now.

site looks good. selection is very limited and you offer nothing that cant be found for comparable prices at places with an established reputation like Dell, or Gateway. and nothing that cant be found for significantly less than discount shops like tiger direct or egg head, or even cheaper on the web.

wouldnt spend alot of money on inventory if i were you.

Where do I go if I want to buy an abacus?

i just bought an abacus for my kid. try any teacher supply store. they are still pretty popular for teaching the basics.

as far as launching a computer store, wow, bad timing. too many people know where to find good prices now. and retail stores like best buy and the shuch are making their money on laptops and the 400 dollar discount boxes. the high end machines are being sold to people who know where to find them for less or can build them.

its a tough market to crack now.

Hi. It's shadowrelm.

You're better off getting 2 dual cores then 1 quad core, and typically half the price.

Then do what, exactly? Duct tape them together? The percentage of the motherboard market that offers more than one CPU socket is about .001. The problem isn't the chip architecture here, and it has nothing to do with Intel specifically. The problem is developers aren't very good at coding for multi-core chips yet. AMD's quad core/three core line is a @#%^ing mess. Their space in the market is completely reliant on power/watt at this point which will keep them alive in the server space, but their share of the gaming/enthusiast market is disintegrating. They need something dramatic not to lose the consumer market completely in the next five years. An architecture breakthrough, or a new fab success. Considering the difficulty they had with the K8 die shrink and the Phenom launch, neither of those seems particularly likely.





Edited, Jul 7th 2008 6:29pm by Smasharoo

www.manofwargaming.com/servlet/the-12/Blaster-AMD-Athlon-3200-pls-/Detail

Get a picture that isn't an empty case and separate your specifications from the flavor text. I could care less if you consider 160GB "spacious", just put it in a pre-formatted table used across the site for quick comparison.

If you're only selling to grandma, ignore the above and focus on your flavor text.

And why do you sell it with keyboard/mouse, but without monitor/speakers? Go full setup or solo tower; giving people half of what they might already have is a waste.

www.manofwargaming.com/servlet/the-Desktops/Categories

Separate your categories and subcategories. I don't need to see the same **** twice--you're just taking up more screen space for no reason.

Have you looked at the new Intel 45NM chips? especially the quad cores with the 12mb L2 cache? low heat, high performance, low energy usage. Intel kicked AMD squarely in the nuts with their new 45NM plant, which AMD is years away from matching at this point, and intel is already working on a 32 (5?) nm die process. I really don't see AMD regaining the gaming market at this point.

Yeah. We get pretty regular dog and pony shows from Intel on their upcoming chip lines. I'm always hesitant to mention some things because I can never remember what's still NDA. They shifted to a new (and darn aggressive) release model a couple years ago that basically has one year as release of a new gen chip, with second year as the refinement of that chip, with a new release the following year. This requires some new "breakthrough" every other year though, which I highly doubt they can maintain through more then a couple cycles at best. I, and most of my peers, kinda assumed it was some execs wish that they release like that, but it seemed awfully convenient to push out the new model right as their 45nm chip was being released with 32nm right around the corner.

Given that it's been about a year and we're still barely seeing market penetration of that chip, I somehow doubt that their 32nm will appear next year, but you never know. They've got designs laid out, and the tech is "doable" (but pretty darn expensive). Going forward, from what I remember from the slides they *have* hit a die size wall past that point. They are basically hoping that some new process appears in the next year or so that'll enable them to go further. I'll also point out that below 32nm you're starting to run into molecular size limitations of the semiconductor strata and oxide layers. And that's just plain hard physics.


I just always take those forward looking deals with a huge grain of salt. I could likely list off about 20 different "groundbreaking and market changing" product lines that any of a half dozen vendors have talked about over the last 10 years that simply never ever materialized. Intel touting this new aggressive release model right as they finally get 45nm to work and with 32nm then possible using mostly the same techniques just seemed like wishful thinking to me in terms of sustainability.

Doesn't mean that they aren't in the catbird seat for a while, but you've got to remember that they're outspending AMD on this like 5 to 1. It's always much much much more expensive to be the guy breaking ground. And Intel has shown historically that it has a hard time fully capitalizing on that, while AMD has shown an ability to step into the space right behind Intel and capture the market when the prices hit the sweet spot. Obviously, history doesn't guarantee the future, but if patterns mean anything, AMD will let Intel take all the lumps and then jump into the 45nm market just in time to grab a good chunk of the broad consumer market at a fraction of the total development cost. Their business model doesn't require that they be the leader, just that while Intel is building the "next new thing(tm)", they're making the current thing that's selling in the market.

The current techniques involve "stacking" layers of chips on top of eachother. But this does produce heat problems. Trying to build a "block" design introduces even more problems though. Heat becomes massively more problematic. You have to remember that a semi-conductor transistor on a chip *is* three dimensional. There are already about 6-8 layers of material alternately deposited and then masked and etched in order to build the chip itself, which does increase its thickness (or decrease and fill depending on how you look at it).

With current fab techniques it's basically impossible to make a single solid state "block" for the simple fact that etching is done essentially 2 dimensionally. So each layer is etched and filled. While I suppose it's possible to do this with a single block of substrate, it's just not practical. The approach more likely used would be stacking separately built layers like a pancake. This could also potentially allow for better heat dissipation, but it's still a problem.


Honestly though, the biggest problem with this is yield rate. On any given wafer, there'll be some number of failed parts. If you do the stacking at the fab, you end up decreasing your yield rate significantly (cause there's X times as many chances to fail for each part). This is why stacking is typically done at the packaging stage instead. Overall yield per package is still kinda poor, but the hoped for is that you gain some benefit doing this. In most cases, you're going to be better off just leaving it as three or four separate chips and then connect them on a board instead (which is how pretty much everyone does it). This allows for greater flexibility, higher yield, easier repair, and overall lower cost to deliver for a final product.


The only uses of stacking I've seen have been temporary. Usually involving multiple designers who cooperate after the fact to combine their chips into a single package, and usually only for as long as it takes to combine them into a single design on a single chip. Even outside of stacking, this is the typical progression.


Obviously, there could be huge benefits to a "block" design, but currently there are a ton of technical hurdles to overcome, and at least right now, they aren't worth tackling. Heck. Packaging alone is a pain. How do you pin a block? How do you then socket the block package onto a board? You have to basically redesign everything surrounding the device in order to do this.

Obviously,

Protip: Everyone's painfully aware that when you use "Obviously, " in a post that you just googled the subject within the last hour or so.

Just linking to the page you're paraphrasing would save us all time.

Er? I'm sorry. I was too busy building offline simulation systems for our new semiconductor design and test lab in Singapore to notice you making a fool out of yourself.