Comments on: AMD Phenom II reviews come out kicking and screaming, but…

By: theovalich

theovalich — Sat, 28 Feb 2009 18:55:03 +0000

Well, in my conversations of late, a topic of 3Way SLI vs QuadSLI came quite often… personally, EVGA GTX285 H2O are the way to go… so yeah, not a huge fan of dual-PCB cards.

By: Car Crazy

Car Crazy — Sat, 28 Feb 2009 18:36:53 +0000

I have had a TRI SLI with BFG GTX280 OC, A Quad SLI with two BFG GTX295, and Finally a TRI SLI with the GTX285 OCX cards and I can tell you first hand the GTX285 OCX stomp the marks of the 295’s. In benches, games, and all other apps involving transcoding etc. GTX285 OCX cards run stock at 702mhz on the core clock. A lot more than that of the 295’s and faster memory etc…. If you have the choice choose the GTX285 OCX in TRI SLI. Or if you have the money try them both like I did. You will decide the same as myself though.

By: Theo Valich

Theo Valich — Tue, 13 Jan 2009 02:04:17 +0000

Unlike Intel, AMD really does not like playing too much with HyperTransport links.

Performance-wise, you can get the highest jump by overclocking the NorthBridge (and just by accident, overclocking the NB memory controller oc’s the L3 cache as well), so I am not exactly

When it comes to my colleagues, I asked around… got answer from some guys, and most had stability issues if they played with it.

Also, AMD sent some motherboards out (I got PII and GigaByte 790GX mobo) to ensure compatibility and maximum overclockability, but… well, my mobo refused to work in Gangbanged dual channel mode. Same happened with ASUS, and updating the BIOS didn’t work.

Thus, the review of Phenom II will happen on the new site, with bits pouring on this blog as well

By: GKL

GKL — Mon, 12 Jan 2009 11:04:36 +0000

Just so you know, some of the retail chips have been hitting 3.6 or higher on stock voltage(air).

Agree with AMD shooting themselves in the foot by sending out old engeneering samples tho lol.

BTW how where you doing the overclock? Almost every review site i seen didnt bother to even raise the HT, they just OC’d using the multiplier which resulted in low clocks and worse performance with those clocks (compared to raising HT too).

By: Theo Valich

Theo Valich — Mon, 12 Jan 2009 08:53:22 +0000

Well, that is true, but it depends on what are you using for cooling… you always want to leave headroom, but after seeing just how my sample of Phenom II was unstable on different clocks, I would think twice about selling anything higher than 3.5 GHz.

After all the hype, AMD managed to drop the PR ball and deliver unstable CPUs, and I haven’t saw any review that passed or touched 4.0 GHz clock on air. If you guys use water-cooling or something like that, you can get it stable – probably.

As far as I go, I have a special way of ensuring stability on machines, and as soon as the site launches and we go with the reviews, we’ll disclose our methods. But the CPU stability method is very simple, and so far, on a spread of C2D, C2Q, i7 and PII, it turned out to be foolproof.

When it comes to seeding press with “early samples”… well, that just sounds like AMD wanted to shoot themselves in the foot. After all the hype, there was a big sense of a baloon exhausting all the hot air…

Of course, I won’t go into the whole “canned benchmark” situation at a certain site that pitted a CPU with 2GB of memory at DDR2-800 clocks against CPUs that had DDR3-1600 memory. Anyways… it seems that press got lame duck CPUs. I won’t file a review that has single-channel DDR2-1066 scores (because neither ASUS nor GigaByte show Gangbanged mode), because quite frankly, system lags behind PI in several benches, and that’s not the experience user will have with retail products.

By: Andrzej Bania

Andrzej Bania — Sun, 11 Jan 2009 10:25:34 +0000

Sorry, let me just add that when I talk about an overclock, I need to differentiate between ‘what we can achieve and leave running benchmarks for >24 hours’ and ‘one that we’d be happy selling and giving a 12 month warranty on’.

For resale, integrated into a system, I think 3.5Ghz or 3.6GHz is more likely – we are investigating this right now ;~)

By: Andrzej Bania

Andrzej Bania — Sun, 11 Jan 2009 10:23:42 +0000

Using a £25 cooler, we have the AMD 940 running stably at 3.75GHz – so maybe the press were seeded ‘early samples’ that were not up to ‘full production spec’.

Also, I notice a lot of BIOS engineers around the globe seem to be claiming overtime ;~)