Intel's marketing - No Celeron

[m25]

Well, the review you mention is not bat, however it's pretty old and the price performance is no quite correct; Semprons are as cheap or cheaper than Celerons and mostly outperform a Celeron of the same price. I will put my hands on and benchmark an AM2 Sempron 2800+(1.6GHz/128K L2) tomorrow; crappy chipset (K8M890) and 512MB of single channel, 533MHz CL4 DDR2.
Currently on Newegg, the cheapest AM2 Sempron (the 2800+) sells for $42 while the cheapest LGA775 CeleronD (326) for $47.
For ultra low builds you had the S754 Sempron 2800+ @ $32 and S478 CeleronD315 for $39.

 

[cwj717]

I got my Sempron working... here are the benches. I couldnt get the Radeon x800 to work in this board though, so i couldnt run the game benchmarks.

Sandra Arithmetic Benchmark:

Dhrystone: 5518 MIPS
Whetstone: 2775 MFLOPS

Sandra Multi-Media Benchmark:

Integer: 11009 it/s
Floating-Point: 12201 it/s

Bryce 6 (Render Time in Minutes:Seconds)

2:39

Tmpg (Time in Minutes:seconds)

1:09

 

[cwj717]

Here are the Graphs

 

[dasickninja]

Hmm, thats odd, the Celeron has a higher Floating Point performance rate than either the Sempron or the P4 2.8...

 

[cwj717]

Yeah, i found that really odd also. Its a fairly old version of sandra... and I also have HT disabled on the P4, that seems to effect sandra score a lot. Plus, I set the RAM timings the same on the P4 and Celeron even though the P4 has much better RAM.

 

[dasickninja]

Try enabling the HT. Just a hunch.

 

[yipsl]

Intel is phasing out it's old outdated NetBurst architecture as fast as it can produce new masks for lower grade C2D's. Pentium 4's will be all but gone by Q3 '07 and Pentium D 8xx series will be gone also. All the information I've seen is that Intel is going to carry its Core architecture over to a new line of Celeron processors.

I expect the single core architecture to be branded as Celerons, but considering Intel's plans, they should ditch single cores completely once quad core becomes mainstream. Then, dual cores with smaller caches should be branded as Celerons.

Vista cries out for a dual core. Will there ever be dual core Celerons at the low end or will Intel ditch the name completely when 8 cores arrive and quad core becomes mainstream?

Celerons aren't a total waste. I couldn't afford both a D865 board, two gigs of RAM, an upgrade to XP and a P4 Northwood, so I bought a 2.6 Northwood core Celeron. A year later, I got a my used P4 Northwood and it's been fine for the past couple of years.

I've stuck with a P4 2.8 Northwood and can't wait till the Prescott, Smithfield, Pressler and other space heater Netburst CPUs are history.

 

[cwj717]

Try enabling the HT. Just a hunch.

I already have the HT scores... I could post the new graphs. HT improves most of the P4s scores (a lot) except the gaming benches... which for some reason take a performance hit. Although, the HT scores also had the better RAM timings. I don't think the RAM timings effect performance that much though.

 

[dasickninja]

Intel's performance has never really been dependent on tight or super low timings. I think the Sempron would benefit more.

 

[m25]

I got my Sempron working... here are the benches. I couldnt get the Radeon x800 to work in this board though, so i couldnt run the game benchmarks.

Sandra Arithmetic Benchmark:

Dhrystone: 5518 MIPS
Whetstone: 2775 MFLOPS

Sandra Multi-Media Benchmark:

Integer: 11009 it/s
Floating-Point: 12201 it/s

Bryce 6 (Render Time in Minutes:Seconds)

2:39

Tmpg (Time in Minutes:seconds)

1:09

Haven't seen that AM2 sempron yet, but Sandra2007 rates a S754 2800+ like this:

Sandra Arithmetic Benchmark:

Dhrystone: 5778 MIPS
Whetstone: 4624 MFLOPS

Sandra Multi-Media Benchmark:

Integer: 15044 it/s
Floating-Point: 16280 it/s

Will come back as soon as I've got some numbers

 

[cwj717]

I got my Sempron working... here are the benches. I couldnt get the Radeon x800 to work in this board though, so i couldnt run the game benchmarks.

Sandra Arithmetic Benchmark:

Dhrystone: 5518 MIPS
Whetstone: 2775 MFLOPS

Sandra Multi-Media Benchmark:

Integer: 11009 it/s
Floating-Point: 12201 it/s

Bryce 6 (Render Time in Minutes:Seconds)

2:39

Tmpg (Time in Minutes:seconds)

1:09

Haven't seen that AM2 sempron yet, but Sandra2007 rates a S754 2800+ like this:

Sandra Arithmetic Benchmark:

Dhrystone: 5778 MIPS
Whetstone: 4624 MFLOPS

Sandra Multi-Media Benchmark:

Integer: 15044 it/s
Floating-Point: 16280 it/s

Will come back as soon as I've got some numbers

Nice. My sempron is the socket A version (aka Barton core, 512k L2)

 

[cwj717]

Intel's performance has never really been dependent on tight or super low timings. I think the Sempron would benefit more.

I would try that with the sempron but its very unstable. The best i can do is get it to boot with CAS 2.5 at 333MHz. I can run the benchmarks at that clock but it always crashes if I open Firefox or run SETI (best it could do was run SETI for 47seconds). Thats with the same ram as the P4 system too. But then its in a chaintech board...

 

[m25]

If that is a barton core, then you can simply find reviews of the corresponding Athlon XP 😀
CL 2.5/333 DDR is like CL3/400, however, not all CL3/400 RRD sticks get well with that timing; I've had the same problems with my old 2GHz Celeron when I OC-ed it to 2.67.

 

[m25]

I'm back cwj717 and others, had a look at the AM2 2800+ and compared to the S754, it had somehow mixed values because the S754 has twice the cache. Also note that the AM2 values are furthe crippled by running it with a x64 system, which makes cache equivalent to 64K. I am putting together a table with all the values mentioned before and mine too:

__________Sem2800+(754)___Celeron2.8G___Sem2800+(AM2)
Dhrystone.......: 5778 ...............5170.................5877 MIPS
Whetstone.......: 4624................3274.................4357 MFLOPS

Integer...........: 15044..............12573................11816 it/s
Floating-Point..: 16280..............13050.................18262 it/s

Also rendered my test scene in Blender; an image of my xV project (http://www.freewebs.com/multid/car/car.html, 512x384px) in ~ 59seconds while my A64 3000+ @1.6GHz runs it in 54.

 

[quantumsheep]

I hate to burst your bubble, but your E6400 is the celeron-equivelant Core 2.

(link)

Don't EVER use Wikipedia in anything that resembles an intellectual arguement, it just makes your point totally invalid.

Wikipedia should die.

 

[dasickninja]

I hate to burst your bubble, but your E6400 is the celeron-equivelant Core 2.

(link)

Don't EVER use Wikipedia in anything that resembles an intellectual arguement, it just makes your point totally invalid.

Wikipedia should die.
Quite true. And what one can prove with a debatable source one can also disprove.

Intel will offer a low-cost single-core version of Conroe, code-named "Conroe-L", starting from the second quarter 2007, according to an article on DailyTech[19]. The new Conroe-L processors will not carry the Core nomenclature. Instead Intel is resuscitating the Pentium and Celeron brands for Conroe-L based products[20].

The New Conroe-L processors shall be named Pentium E2xxx series. It supports 800MHz FSB and only has 1MB L2 Cache.

I don't see any mention of the Conroe "Allendale" core being a Celeron.

 

[quantumsheep]

I hate to burst your bubble, but your E6400 is the celeron-equivelant Core 2.

(link)

Don't EVER use Wikipedia in anything that resembles an intellectual arguement, it just makes your point totally invalid.

Wikipedia should die.
Quite true. And what one can prove with a debatable source one can also disprove.

Intel will offer a low-cost single-core version of Conroe, code-named "Conroe-L", starting from the second quarter 2007, according to an article on DailyTech[19]. The new Conroe-L processors will not carry the Core nomenclature. Instead Intel is resuscitating the Pentium and Celeron brands for Conroe-L based products[20].

The New Conroe-L processors shall be named Pentium E2xxx series. It supports 800MHz FSB and only has 1MB L2 Cache.

I don't see any mention of the Conroe "Allendale" core being a Celeron.

You truely are a man after my own heart Ninja

 

[dasickninja]

Why thank you.

 

[4745454b]

Intel will offer a low-cost single-core version of Conroe, code-named "Conroe-L", starting from the second quarter 2007, according to an article on DailyTech[19]. The new Conroe-L processors will not carry the Core nomenclature. Instead Intel is resuscitating the Pentium and Celeron brands for Conroe-L based products[20].

The New Conroe-L processors shall be named Pentium E2xxx series. It supports 800MHz FSB and only has 1MB L2 Cache.

I don't see any mention of the Conroe "Allendale" core being a Celeron.

Because they aren't. No current Core/Conroe offerings are celeron chips. Good luck convincing some people around here of that...

 

[dasickninja]

If anything it can be compared as a P4 3.2GHz to a 3.8GHz.

 

[Whizzard9992]

First off, I'm not insulting anyone's processor, so come people need to stop taking this so personally. I was merely trying to provide a different perspective into the recent events surrounding the release of Intel's Core 2 Architecture.

Let's focus on the point I'm trying to make, and not defending the title of the Allendale. I don't think that they should have named them Celerons, nor did I ever say they should have. I said that if Intel continued with their tradition of labelling cache-failed chips as celerons, then the E6300/E6400's would have been Celerons. They didn't, and that was a marketing decision.


The Allendale is actually just a Conroe with 1/2 the cache disabled.

Intel's Code Names for Allendale
Cool and cheap screamers from Intel and AMD

You'll find plenty of sources that will tell you the code name for these 2MB Core 2 Duo processors is "Allendale," but Intel says otherwise. These CPUs are still code-named "Conroe," which makes sense since they're the same physical chips with half of their L2 cache disabled. Intel may well be cooking up a chip code-named Allendale with 2MB of L2 cache natively, but this is not that chip.

The old Celerons were just P4's with 1/2 the cache disabled, and various features disabled.

Instead of designing an entirely new core for the Celeron, Intel simply disabled 128K of the Coppermine's 256K L2 cache and then called it Celeron. (link)

Apparently, based on the article in www.overclockers.com.tw, a Celeron just has half the cache disabled via a control line. (link)

The reason the Celeron has more is the fact that its really a PIII with half the cache disabled because of a fault (link)


Look for yourself.... (link)

All Intel had to do, if they chose to, was to disable SpeedStep, VT, or w/e, and lable them celerons, the same way they did with the NetBurst chips. That's all I'm saying. I don't really care about the celeron's performance. Forget I said anything about that. It was just an opinion I stated and a couple people latched on to for some stupid reason. It's obviously an arguable point and has nothing to do with the original post.


This time, if you want to refute my sources, prove to me that Celerons (P3 and later) were completely different processors, and not just Pentiums with disabled caches and features :!: