Thursday, March 25, 2010

AMD's hope...



Last post I basically lamented my lost hope for AMD's platform. Well, there still may be hope, if this past week's news is to be believed. First, is the news that AMD is getting good results internally with their SATA 6G controller, so if that can be translated to this external world of ours, that gives them a competitive edge finally. Second, X-bit's update on the status of the new CPU's implies we should see them any day now. It's not the Thubans that are so interesting, though, it's the Zosma. It has more L3 than the current quad-core Phenoms, so assuming latency hasn't been shot way up, performance/clock should improve a little. Combine that with 400Mhz turbo boost, and a (hopefully) low price, and we may actually have a competitor to the i5-750. Especially considering Intel hasn't moved their quads to 32nm yet.

Both of these hopes need to be realized together. Please AMD, for the sake of yourself, don't miss this opportunity.

Wednesday, March 17, 2010

Why does Intel need an ICH11 if others can't beat their ICH10?

I was hopeful for Nvidia’s Ion 2. Perhaps it was going to re-align itself with the price drops on low-voltage Core 2 CPUs, making Macbook-like Windows laptops available with awesome battery life and performance for under $600. Perhaps it was going to have a graphics upgrade, to 32 shader cores, and thus making a desktop platform where the integrated video could be re-purposed for dedicated Physx. But nay, it’s merely an attempt to toe in some Nvidia product into an otherwise standalone platform.

Intel has been on a rampage privatizing their platforms lately, and all that’s left, sadly, it the Core 2. Since the Clarksdale’s “chipset” is just on package and not on die, Intel could easily make a Clarkdale for socket 775, but I wouldn’t hold my breath for that, as that would kill sales for much of the existing midrange Core 2 inventory, as well as giving room for Nvidia to compete again.

But within Intel’s drive to bring the entire platform under their wings, lies an opportunity… unfortunately, I think both AMD and Nvidia missed that opportunity. Intel is still sitting on their ICH10 “southbridge”. The power management technologies of SATA 6G and the sheer performance increase of USB 3.0 are not inherent in Intel’s platforms. These are both features that need to be added separately, at additional cost and power consumption. Intel has yet to release a low voltage Arrandale: imagine if the already impressive Macbook Pro got another hour of battery life and had USB3 support… that’s what Nvidia could have done with ION 2, but instead it’s merely a graphics upgrade to a platform that really isn’t meant for it. And AMD? They just released their new desktop platform, and while it has SATA 6G, benchmarks show it still falling behind Intel’s controller with contemporary drives, especially the fastest SSDs. At this point, for a desktop, USB 3.0 is the more important upgrade, whereas the power management upgrades of SATA 6G would be more beneficial for laptops. It’s like AMD got it backwards.

I really wanted to see the latest platform updates from Nvidia and AMD give some sort of competitive advantage in some area: if you have a CPU disadvantage, then you should strive like mad to create an advantage in the rest of the platform. If the new SATA and (the hoped for) USB aspects performed well , I would have seriously considered building my next system based on a Phenom II, simply because a overclocked quadcore Phenom II is not going to be bottlenecking the apps I use. But right now, outside of reaching as low as price as possible, I see absolutely no advantage in AMD’s platform whatsoever… and it’s saddening because it didn’t have to be that way. So, it looks like the two biggest aftermarket GPU manufacturers only have GPUs to talk about for the near future.

Thursday, March 11, 2010

Asus RT-N16

I haven't had much luck with routers since I "upgraded" from my uber Dlink gaming router a couple years ago. I wanted an N router that reached to every corner of the house. The Belkin N+ was supposedly one of the best. It seemed to be a good router for awhile, until I noticed it was dropping my internet connection to all connected clients every so often (less than an hour between dropouts, on average) for a few seconds. Even if there was only one computer connected via ethernet, and it was in the DMZ... same problem. Tried different firmware, and even exchanged for a newer revision (from v1000 to v2000): same problem. Apparently a v3000 is out now, so might try exchanging it again, but my new router for now is the Asus RT-N16. It initially seemed fine, albeit less throughput than the Belkin, but there seems to be some QOS issues in Asus's firmware. After a week of use, it would assign all resources to just one download, regardless of how I configured the QOS settings: if I was downloading anything, I couldn't open a page or start another download. Since Asus was still sitting on their release firmware, I tried DD-WRT. It works. Wonderfully. The RT-N16 still doesn't have the raw throughput of the Belkin, but the QOS works beautifully: torrents and gaming can exist side-by-side now!

I didn't want to post any positive news until it had proved itself, but it's been around a month since the DD-WRT flash, so I feel I can safely say it's recommended now.

Monday, March 1, 2010

Asus U3S6 USB + SATA PCIe card

Asus has a list of boards that the card is compatible with, but users are reporting much wider compatibility. Well, I tried it on my Asus X48 board that had a spare 16x PCIe 2.0 slot open... and no dice on the USB 3.0 (the SATA controller seemed to install though).

I then tried it on an old P35 board with a "16x" slot that was electrically a 4x PCIe 1.1 slot, and it worked fine.

So, I can't say for sure it doesn't work in PCIe 2.0 slots... but something to keep in mind.

*Update* I looks like it was due to an overclocked PCIe bus. I ran into the same problem with the Gigabyte 3.0 card on the same board. I kept scaling the PCIe speed back until 103Mhz, at which point the NEC USB3.0 controller sprang to life. So, keep in mind, the NEC USB controller won't take higher than 103 MHz on the PCIe bus, at least from what I've seen.