Slashdot Mirror
AMD Says Barcelona Will Outperform Clovertown
Dysfnctnl85 points out a ZDNet Blog posting in which AMD claims that its upcoming quad-core "Barcelona" chipset should be 40% faster than "Clovertown," Intel's quad-core Xeon 5300 line. AMD says that the introduction of Barcelona marks a shift in their strategy from emphasizing price to performance. The post goes on: "Intel is eager to claw back some of the server market share from AMD, and this is where Clovertown comes in... The Xeon 5300 line will represent excellent value for money since Intel plans on pricing them the same as its dual core Xeon 5100 processors. That could make things tough for AMD."
Yes, there were four cores. Two cores on one chip, two cores on the other. They key words from the titles you quoted are "platform" and "multi-socket", neither of which imply a single die with four cores.
A deep unwavering belief is a sure sign you're missing something...
Latency is not a bottleneck.
You could turn it around and say that, since the disks are not using their full bandwidth, the disks spend most of their time waiting for requests.
That being said, disk latency is one of the major causes of poor performance. But "bottlenecks" only have to do with throughput.
Social scientists are inspired by theories; scientists are humbled by facts.
Check the subject line. Clovertown and Barcelona are codenames for CPUs, not chipsets. "Chipset" is used to refer to the chips used to support the CPU, not the CPU itself.
This 40% faster than Clovertown claim is only referring to FP code. The integer side is not nearly as clear. Expect AMD to improve integer performance over K8, but I don't expect any miracles. Here is a small list of improvements Barcelona will have over K8:
- Double L1 cache bandwidth
- Double FP units
- Single-cycle SSE (vs K8's 2-cycle)
- More fast-path decoding
- Double TLB size
- Independent DDR channels
- More cache (L3)
- Out-of-Order loads
- New instructions (LZCNT, POPCNT, EXTRQ/INSERTQ, MOVNTSD/MOVNTSS)
- Double prefetch (from 16 bytes -> 32 bytes)
- Larger Branch Target Buffer
- Larger Out of Order (OoO) buffers
- Support for new HT standard (3.0)
I am a viral sig. Please help me spread.
Urm, that depends. Linux CAN be a pre-emptible kernel, if you compile it to be. There are various levels of pre-emptibility, depending on your needs. The in-kernel docs say that pre-emption is intended for desktop environments where perceived latency is a big deal, but servers will probably benefit from the lessened overhead of a non-pre-emptible configuration.
But the original poster's comment is still bullshit. Windows Vista is a microkernel? What has THAT guy been smoking? Multi-core designs aren't that different from multi-CPU configurations, and we already know from experience that Linux hasn't been sidelined performance-wise.
Actually, now that I think about it, the likeliest explanation is that the OP was just trolling.
Then you must be new. AMD has basically been outperforming Intel ever since the 1.13Ghz PIII debacle. By Thunderbird and Thoroughbred in particular, AMD was outperforming the PIVs at about half the price. It wasn't until AMD64 that AMD started charging Intel prices. That's at least four years where AMD equaled or beat Intel performance for drastically less money (on the order of 50% price and often times less).
A SATA 1 interface can transfer at a maximum of 150 megs/s, but your hard drive can't. On sequential reads, you're unlikely to see much higher than 40 megs/s, even 7200 RPM desktop drives don't exceed 70 megs/s yet.
Game! - Where the stick is mightier than the sword!
Since you were very snotty to someone who replied to you, I'm posting the snotty version of my reply... There is a point at which one must face facts. Since the AMD 64 line, AMD has had the superior cpu over the P4. The AMD line was, with few exceptions, cheaper, cooler, faster, and as stable, if not more so, and it did that while still including your precious protection against forgetting to put on the heatsink. Is that something you do often? Yes, the Athlon line was neck-and-neck with the P3, the Athlon XP line did well against the P4 until the end, the 64 line, like I mentioned, kicked ass until the Core 2 Duo came out, something you might not be aware of since you're posting problems with AMD cpus that haven't been an issue since 2001. Yes, the Core 2 line is awesome, but for high-end server work (expensive stuff), the AMD architecture still has some very nice things going for it.
RAID 0 doesn't automatically ensure that both/all drives will read/write at the same time. In fact, RAID 0 is typically configured NOT to optimize that. Large interleave sizes, the ones typically used, optimize for multiple concurrent requests, not parallel throughput. Furthermore, each drive in a SATA system will get its own SATA link, so unless you are attaching a RAID 0 box via a single SATA link, your entire point is specious. Even if you had 2 RAID 0 drives with a small interleave attached via a single SATA 150 link, you've proven that 150 MB/sec is fast enough.
AMD has yet to release a quad-core CPU. The whole QuadFX(aka 4x4) is a dual-processor system that is designed to be a consumer level platform that supports two processors. The reason for QuadFX to be worth looking at as a platform is that you will be able to swap out the two dual-core processors for two quad-core processors starting around the middle of 2007.
AMD is really starting to hype their true next generation core design, not just quad-core. This is something that many people seem to be closing their eyes to about this particular issue. The current AMD processors are based on the AMD K8 design(regardless of fab process, amounts of cache, or tweaks to the design). The new stuff being released this summer is based on what is called the K8L design. Yes, it's still the same basic design as the current K8, but you could say that Intel's Core designs are based on the Pentium-M, which was based on the Pentium 3, which was based on the Pentium 2, which was based on the Pentium Pro. Obviously there will be similarities here to previous processors, but the progress between the last implementation from either company should not be ignored.
The Pentium-M was a great laptop processor, but if you tried to make it compete without a LOT of changes with the desktop processors from AMD or Intel at the time, it would NOT have done well on the desktop except in the low power usage area. So, how many changes did Intel make to turn it into the Core line of processors?
The Athlon 64 is a great processor, but AMD knows that it doesn't compete well against the Core line of chips. So, they are doing to it what Intel did to the Pentium-M, do a HUGE overhaul of the design. Intel hyped their Core chips well before release, even while the Pentium 4 was being beaten left and right by the Athlon 64, so now the shoe is on the other foot. If the new K8L design is as good as the hype indicates, then not only will the AMD quad-core beat Intel's quad-core by a good amount, but dual-core processors based on the same design should also compete very well, though probably not quite as well as the quad-core.
The reason why quad-core might compete better is based on the connections between the CPU and memory. Remember that Intel is still using the CPU-chipset-memory design while AMD connects the CPU to the memory directly. As a result, when you go from dual-core to quad-core on a native design, Intel processors will still need to go through the chipset to get to the memory. Unless the bus speed is really increased to provide that extra bandwidth, the Intel processors will end up starved for memory. In fact, that's a part of why you see so much cache on Intel processors currently, to compensate for the old design that Intel is still married to.
Intel still hasn't released a native quad-core implementation btw, they need to go through the system bus to connect each pair of cores on their "quad-core" processors.