AlexHopmann.com

Of course it looks like I jumped the gun a bit in some of my conclusions in my previous post about video cards. ATI also announced their new generation of cards, the HD 3800 and they look pretty impressive too. Of course at the moment it looks like a bit of a vaporware launch (to head off today’s NVidia announcements), with no benchmarks / details, but the real launch is less than 3 weeks away so it seems like a good time to wait a bit before deciding.

Lots of hardware news is out today. First of all an update on an older item. I bought the Gigabyte SilentPipe GF 8600GTS for the Media box a few months ago. I guess its not the end of the world since I did need something until now, but it turns out to have been a mistake. The Zalman HD135 case supports full height cards, but just barely. There is no room over the top of the connector but the heat sink on the SilentPipe card goes above the top and the top of the case doesn’t fit (although its very close).

Also this is hardly a totally silent case anyway- with the thing buttoned up I had some heat problems so I probably need better cooling on the video card and/or another fan anyway.

I’d been thinking about maybe picking up one of the AMD (ATI) Radeon HD2600 cards. They have great media benchmarks and I’m not really planning on using this machine for 3d. But today along comes details of the NVidia 8800GT which should be faster than the existing $399 8800GTS, and available for just about $200. Its built with a better process (65nm) so its lower power, and there are 1-slot versions so it won’t block the slot next to it. Hopefully even a normal fan-cooled version of these won’t be too loud, and I’m probably going to want the same thing in my workstation too (since it blows away everything in the AMD line for gaming and only the $600 card is any better and that isn’t by much).

Of course this card (especially at $200) makes the motherboard choice even more painful since using NVidia SLI starts to sound reasonable. But the Intel P35, X38 and X48 chipsets don’t support SLI, so you are left with some pretty painful choices.

On the processor front, it does appear that Intel is going to have an upgrade to the 3.0ghz 1333mhz bus Penryn chip in Q1 2008. The upgrade should be 3.2ghz with a 1600mhz bus. The clock is only 7% faster, but its one of those things where if you combined it with the faster bus AND fast DDR3 memory the whole thing might be good for a 20% improvement. But even with DDR3 prices coming down, they are still likely to be 2x DD2 in 3-4 months and its hard to imagine that you don’t end up paying $600 extra for that 20% improvement.

I wonder when NVidia is going to come out with a new chipset? Their 680SLI is feeling a bit old compared to the new Intel ones and given that they have a video-card coming out with PCI 2.0 I’d expect them to do a chipset to go with it.

Does anyone know if there are touch LCD displays that can sense more than one touch at a time? I’d love to experiment with some multi-touch things around the house but not sure where to get an appropriate display.

Ideally 20″ or larger although if its really expensive I can’t do it.

One big problem that technology hasn’t solved at all. I’ve got tons of cookbooks. They fill an entire cupboard in my kitchen. Yet, when I want to find some good recipes for something specific, it can be a total mess. I want to cook a killer cassoulet. I love cassoulet, and its one of those dishes that can be a great test for a place that tries to make it. But of the 100s of cookbooks, where can I find great cassoulet?

I eventually found what looks like a good recipe in the Balthazar cookbook, but its still a mess. Often for recipes I wind up going with the Joy of Cooking or Larousse instead of something more interesting just because they have everything. I’d kill for something that makes it easier to find what I want…

Slate has an article on the economic analysis of whether newspapers should charge for their online versions. Now before I go on I should say that us Internet types are supposed to pooh-pooh these kind of analysis as “old school” and “not understanding how the Internet changes things”.

So I’ll get right on it. Don’t want to disappoint after all.
t
My reading of this article is that all its analysis is in terms of traditional models. It asks the classic economic questions “does one product (the online version) act as a replacement for another product (the print version)”. And it looks at pricing/revenue models in terms of subscription vs. advertising. It suggests the big thing that has changed is that advertising on the Internet is becoming more mature and thus there is real money in it.

Of course the article totally ignores the ways in which the Internet is not like other media. Their analysis might be appropriate for comparing newspapers to TV news (with TV having free + advertising vs. subscription, not that anyone was crazy enough to try to charge for the subscription beyond the inclusion in the basic cable package).

What’s missing about the analysis of the Internet sites is a discussion of the effects of linking and search engines, and those don’t really have a comparison outside this new medium. By keeping lots of their content locked up inside a subscription site the NY Times and Wall Street Journal have by and large kept their content out-of-play from the rest of the Internet.

When looking at any Internet property, you consider three types of non-paid traffic sources- direct, referral, and search. Sure, there are some web-sites that people will regularly go to on a frequent basis just by typing their address into their browser, and relatively speaking a daily newspaper is a good candidate for this since their most important content has a low shelf-life and thus if you are prominent enough you will get a fair amount of direct traffic. But even among daily news a large amount of traffic can be from blogs, link blogs, news aggregation (like Google News), and search. These publications actually do tons of articles (food reviews, recipes, movie info, and more) that would be great targets for organic search and could generate substantial long term advertising revenue if they were available.

The NY Times had the worst of these worlds. Much of their daily content was available for free, but after a week or two it would get locked away. So they didn’t get the ability to really monetize that old stuff and people would be reluctant to link to their news since the links would go dead after a bit.

Hopefully with the recent changes these publications will actually join the web (the world of interconnected sites) and I’m expecting with the huge value of their content they will be able to make some great bucks off that.

Fun, I think I just got my first push poll. Whats especially surprising is that it seems like the Sierra Club or their allies are the ones paying for what is considered a fairly slimy tactic.

I received a call that asked me to spend 4 minutes participating in a poll that wouldn’t need my name. Why not I figure? The first questions is pretty simple, how do you plan on voting on the ballot issue #1 (the levy for road + transit construction in the Puget Sound area). I press something.

Next they say something about how “did you know that this measure is opposed by the Sierra Club because its a bad solution to Washington’s traffic problems and creates more global warming? Knowing this how would you vote?” At this point its clearly not trying to get any real scientific data- rather in the phrasing of the question they are trying to state certain controversial issues as facts. They went on to ask another couple of questions, each really unloading dumps of toxic sewage on the ballot measure.

Hmmm. With this kind of slimy tactic, it sure makes me want to vote for it.

Joel writes on outsmarting your airline. I do this all the time- use FlightAware and other services to tell when your incoming flight is actually arriving. Also as a pilot I have access to aviation weather and flow control information and I’ve often checked up on them when the airlines claim a flight is delayed due to ATC flow-control or weather. I’m estimate is that for the major airlines like American, United and Northwest, about half the time they claim delays due to ATC or weather its a lie. Which is not to say that the gate agents are telling a lie themselves- they may not know, but someone at the airline is. More often than not the cause is bone-headed airlines that have scheduled 50 of their own flights to take off or land at the same moment at one of their hubs, which is just inexcusably bad management.

An interesting contrast in airlines is our trip out to Minnesota two weeks ago. Outbound we flew Northwest. I just missed taking a photo of the gate sign showing our “departure time” as 10 minutes in the past. The flight was late (ok, it happens, no big deal) but they never bothered to try to keep people informed or update the displays anywhere.

Our return flight was on Sun Country, one of the newer-style airlines. The contrast was amazing- the return flight was also late. But they made announcements every 10-15 minutes keeping us updated on the status, they frequently updated the board with a projected (and fairly realistic) new boarding and departure time, and even suggested that people call ahead if they having someone picking them up to warn about the potential delay. It was the model of good customer contact through proactive honest communication.

I just don’t believe that its that hard for an airline to keep passengers and gate crews informed of what is really going on- heck, I was in Bend Oregon the other day being picked up by a friend in a small Columbia aircraft and could track his arrival with better precision than Northwest or the other guys would ever communicate to their customers. Travel can be complicated and stressful and if these big airlines would spend 10% of what they spend creating ads showing the beautiful travel experiences, on actually giving a better experience, people might not hate them so much. Instead they are just lazy and take the short cut of telling deliberate lies to try to placate their captive audience. Lets be clear- if it were not for government subsidies, the old guys would have already been out of business by now in a classic case of good capitalist “creative destruction” and there would be more room for the new guys like Sun Country, JetBlue and Southwest to replace them with good service.

Dana wrote a great piece on Perfecting Panna Cotta over on Tastingmenu.com. I think some people had a bit of a tough time initially figuring out what to make of the “new” tasting menu, but with pieces like this Dana is really putting it over the top- this article I’m sure will be one of the primary resources on the Internet for how to make Panna Cotta. And I don’t mean just for a couple of years- its really cool to see new content that will be useful for decades.

One note- she mentions that its hard for the home chef to find gelatin sheets and I’ve found that to be very true. Last time I needed some I managed to borrow them from a pro kitchen. But today we found some gelatin sheets at Delaurenti’s down in the Seattle Pike Market.

Some more data is coming out on the real performance characteristics of DDR3 memory. The good news is that it looks like we are going to see some huge increases of memory bandwidth from the typical 800mhz today to 2000mhz soon (the over-clocking potential of some of the new memory is a good sign of what typical memory will do in 2 to 3 years).

The bad news is that as typical for high performance computing, the impact on overall system performance is still small. Computational/memory intensive benchmarks are showing a 2-3% improvement going from DDR2-800mhz to DDR3-1333mhz (which is the “normal” DDR3 speed now) and maybe 5% improvement from DDR2-800mhz to an over-clocked DDR3-2000mhz. I’m sure it would be different with a 5ghz processor or more likely if you are running 8-cores instead of the 2-cores they used for testing. This is one of those typical situations where performance improvements to one component only have a small impact unless that specific component is really holding back the rest of the system. Try comparing a 3ghz quad-core CPU with 400mhz memory vs 800mhz memory if you want to see a much more direct relationship between the memory speed and overall system performance.

Meanwhile, DDR3 memory is still psycho expensive on the range of $400 for 1gbx2, vs around $70 for the same 2gb in high quality DDR2-800 or maybe $130 if you want to get DDR2-1066 sticks (although the enthusiast sites suggest that most DDR2-800 chips run at DDR2-1066 easily).

The catch is you have to bet on a memory technology when you buy the motherboard and presumably within a year (and maybe much sooner) DDR3 will become mainstream and much more reasonable. Still considering that I’m interested in at least 4gb for my workstation, there is no way I’m paying an extra $700 for RAM. Also there is no such thing as 2gb DDR3 DIMMs so if I want 2gb on each stick (for future expansion room) I need to go with DDR2 anyway. 4gb (2×2gb) DDR2-800 is only about $160, or for $190 I can go for DDR2-1000. Since the newer Asus boards all officially support 1066mhz memory speeds that seems like a pretty safe bet to squeeze a little extra performance out for only a few extra $.

The other news out recently is that ATI should be shipping a new generation video board based on a die-shrink soon. The interesting aspect of this news for me is that they might have a new board out based on the RV670 that will have lower power consumption and only take up one slot for the cooler, not blocking the slots next to the video card, yet it will still have similar performance to the current top of the line Radeon HD 2900s. Of course on my workstation I don’t really have anything I was planning in putting in any of the other slots, but its nice to have the extra flexibility.

Got the link in my email this morning. The download went flawlessly, all the concerns about Radiohead being able to deal with the infrastructure seemed unfounded. To be honest, it went more smoothly than my Amazon mp3 download experience since Radiohead just sent out customized links and didn’t try to do some wacky “you can only download once and you are screwed if the download fails” sort of thing. Its amazing how people can just make things more complicated for themselves (and their customers!) in unnecessary ways… I wonder if Amazon should try to block TCP-level retransmits to make sure you receive each packet only once?

Two things that could have been better- I wish the files were encoded at more than 160kbps. Its better than 128kbps but I’d rather see 256-320kbps to get much closer to lossless quality. I would have expected that with the band managing the process they would have cared about making sure their “product” was at the best quality possible, but then again most people won’t notice or care and I suppose it does preserve some value for people who buy the physical CD.

I also wish they had included some album-art files. Its nice to have them in Explorer / the Media Library and too bad they didn’t just include a few of the correctly named JPG files.

The music? Its almost beside the point right? Listening to it now. In any case the last few Radiohead albums took me a few play-throughs to get into so I’ll wait a week or so before making up my mind.