Prescott not faster than P4s?

[G|-|oST]

Here

Any thoughts on this? Any people have some inside info?

[NaHeMiA]

I heard that it is 10% faster then the P4C. The EE is still top notch though

[Maelstorm]

According to that article, it is saying that they predict that the initial Prescotts will be hardly faster than the Northwoods when it comes to gaming. They predict that Media applications will see the most gain.

I say that only time will tell.

[NaHeMiA]

I didnt read that article but I remember reading one from Xbit Labs a few weeks ago.

[DazzXP]

IT WILL INDEED BE FASTER THEN A P4 OF THE SAME SPEED. Why just the extra level 1 & level 2 cache will boast performance quite a bit, P4 likes the extra cache for it's high latency memory to be stored.

[Sphinx]

Prescott has more latency on cache L2 than Northwood, that is also one problem!

[GRAFiZ]

It doesnt really matter if it isnt any faster then the current P4 at the same clock speed.

The more important fact is the improved core design and higher clock speeds.

Lets just say the Prescott performs exactly the same as a P4c at 3.2ghz.

Fine.

But when the Prescott is at 3.8ghz or 4+ thats when it will make sense.

It's not a P5... it's just an revision of the P4.

Get it for the clock speed.

[DazzXP]

From the preview i saw in most cases a Presctott 2.8Ghz was as fast as a 3~3.2GHz P4.

[Soulburner]

Originally posted by DazzXP
From the preview i saw in most cases a Presctott 2.8Ghz was as fast as a 3~3.2GHz P4.

Yes this is what I saw as well.

Anyone saying it won't be better than the Northwood is full of it.

[DazzXP]

But it did run pretty hot too, alot hotter then a northwood, however that is to be expected due to the smaller die which will given less contact area.

[Sphinx]

Originally posted by DazzXP
But it did run pretty hot too, alot hotter then a northwood, however that is to be expected due to the smaller die which will given less contact area.

Don't worry about that, it wan't be too hot under MachII!

[skate2snow]

sorry, but thats only some preview, not some review, the only thing we can fix on is the review. I think preview should not say anything of faster or slower, coze the only thing they can say is what it will be and even theyre there not sure. Thats my opinion

[lalPOOO]

Personally I think the article carried more good news then bad. The larger cache sizes and the other improvements look quite nice. The part about it being slower clock for clock, or on par is one persons speculation which is worth taking into account, but its not the truth. If you look, theres usually tons of threads like this all over the internet before a new product is released, and in the end they all end up being showed up. Example? "9800 pro is just an overclocked 9700 pro, theres almost no difference" or all that nonsense about how the 3.2EE barly overclocked and was just a big expensive peice of garbage. You just gotta have a little faith, generally these companies don't stuff worse then they've already made. (please don't bring up nvidia.)

[Soulburner]

Exactly like I said before...

Look at what everyone expected out of the A64...and look how it performs after everything thought it wouldn't be anything special.

[skate2snow]

everybody panic before to see the true results

[SpudHo]

Here's what Ace's Hardware had to say (best review site on the web, I think).

"Prescott is basically a Northwood Pentium 4 with the following improvements:


bigger D-L1 cache (16 KB instead of 8 KB) & L2-cache (1 MB instead of 512 KB) .. No comments necessary

4x Improved Clock Distribution (compared to Northwood) for better Frequency Scaling

Automated design of the functional blocks for better clockscaling

Improved Imul latency : Northwood/Willamette do their integer multiplications on the FPU, and the big latency is due routing the data between integer and FP datapaths. Prescott has a dedicated integer multiplier. (Thanks goes to Heikki Kultala).

Prescott New Instructions (SSE-3), which will not improve performance at the launch (needs optimized software).

Additional WC Buffers. Instead of sending small pieces of data to the AGP videocard, these pieces of data are stored together in buffers, and send through in one big burst. This helps to preserve FSB bandwidth as the bandwidth of the FSB is more efficiently used (less overhead from one big burst than from many small ones)

Improved Pre-Fetcher Branch Predictor. I did not get much info on this but it seems that the buffers have been made bigger so the branch predictor will be able to cope better with more than one thread.

Improved Hyperthreading: two new instructions: Monitor and Wait, which will only improve performance on recompiled software.

we can add two more, one being a fact and another being a rumor:

longer pipeline (fact)
Higher L2-cache latency ? (rumor)"


edit: readability

[DazzXP]

Yeah but the Net burst architecture really does use the extra cache unlike the short pipeline on the Athlon. Remember the speed boost from the Willamette to the Northwood, only thing done at the time was and extra 256k cache. Now they are doubling the 2md level cache once again as well as doubling the 1st level cache.

[drunkenmaster]

yes, but l1 cache is useless when sizes are larger than it, which is most of the time as its tiny, 128 on a xp is far more effective.

Soulburner, i don't know if anyone, and certainly not me, thought the ath 64 was gonna be nothing special, infact i've been on the edge of my seat waiting for one for a year and a half, and its as good as i expected.

THe other improvements are fairly uselessm extra l2 is the only useful thing and if its true and it runs at a latency penalty then the whole lot will work less effectively.

Anyway, i expect it to be very much cache improvement speed boosts only, what intel says about it pretty much supports that and its likely to be 5% faster max. There is no way in sweet hell that a 2.8Ghz prescott will beat the 3.2Ghz p4, no way in hell. In maybe one specific benchmark that really depends on cache size of course there may appear to be that kinda speed diff. but gaming and a general overview it won't be much faster.,

AS has been mentioned the only advantage of the prescott is clock speed, but, its not going well. early last year the idea was prescott a few months ago scaling to 5Ghz by the end of this year, the roadmap has changed to show the prescott only at 4Ghz by the end of the year, and then tejas hitting. SO the prescott is hitting a full Ghz less than Intel wanted, AND, tbh, the p4 normal as it is now could have been tweaked to 4Ghz. Its all looking bad, and for now, 0.09 is not working for them so good.

But that just means they won't be happy with stock speeds that high, they might clock up very well just with too much heat for stock sinks. So hopefully we'll see good clocks, we'll see soon enough.

[STEvil]

If I am not mistaken, Prescot has no 64bit capabilities, either, correct?

EDIT

A mention about the memory controller onboard the AMD's, also.. does it really matter that its onboard whether the ram is DDR or QDR?

I mean, all the memory controller has to do is tell which bank for the data to go to, and its likely that AMD already knew QDR was comming and built in the ability to work with DDR or QDR into the chip... if it needs that to be built in at all.

You might as well build a new CPU if you are going to rework it internally to the extent that changing a memory controller may require..

[Soulburner]

Originally posted by STEvil
If I am not mistaken, Prescot has no 64bit capabilities, either, correct?

A lot of places report it does infact have 64 bit instructions, however they won't be "active" yet, just like the P4 Northwoods had HT a long time ago (look at the pins, haven't changed a bit since 1.6a) but it wasn't "activated" until recently.

[hollywood]

Originally posted by skate2snow
sorry, but thats only some preview, not some review, the only thing we can fix on is the review. I think preview should not say anything of faster or slower, coze the only thing they can say is what it will be and even theyre there not sure. Thats my opinion

A well thought out point Skate!!!

[st0nedpenguin]

Just think back to the release of the P4, it was trounced by the P3 performance wise, but once motherboard and software support had been optimised it started slowly gathering speed.

Confucious had a famous quote relating to the performance of new Intel CPU's if I remember correctly:

"Shudduppa your face, it's not even out yet!"

[drunkenmaster]

no, the p3 still trounces the p4, easily , on clock for clock. THe p4 didn't get better, simply faster.

Also, performance of new cpu's, compared to 5 months later doesn't change ,it can do what it can do, no more no less. THe ath 64 might "appear" faster with 64 bit software, but, we don't know how fast it is, in 5 months the ath 64 will still be the same speed in 32bit apps as it is now. They don't change the architechture of a cpu over time.

Software may become more efficient over time but thats it. INtels general plan so far has been decrease IPC to ramp up clock speed. Though, i think this time although they might possibly do the same overall with the core, smacking in the extra cache makes up for it and makes it slightly faster.

[Soulburner]

But I don't think they lowered IPC this time.

[st0nedpenguin]

Improvements in supporting chipsets and motherboards can easily increase the performance of a new CPU, the same with software optimisations, especially if you're looking at a new set of instructions (as long as those instructions aren't MMX ), admittedly, the chip itself isn't actually running faster, but it helps to eliminate any bottlenecking caused by inferiour support on release.

Just as an example, slap an AthlonXP in a via KT400, and then in an nForce2, I'm pretty certain it'll perform better.