AMD A4-5000 vs. Intel T9500

[Turbonium]

I've been looking at some benchmarks online (keeping in mind benchmarks are often overly synthetic).

In the real-world, is the AMD A4-5000 truly as good as the Intel Core 2 Duo T9500 (give or take)?

I'm talking in terms of raw CPU performance here, and ignoring power consumption, cost, or other metrics.

 

[frozentundra123456]

They have very similar passmark scores, which is surprising to me, but the A4 is 4 cores at 1.5ghz, and the Core 2 is 2.6ghz. I dont think the ipc of Kabini is faster than even Core 2 Duo, so in normal use, lightly threaded apps, I think the Core 2 would be faster, while the A4 might be equal in perfectly threaded loads. The only data in Anand's bench was for Cinebench R10. I made the closest comparison I could, E4700 at 2.6ghz vs Athlon 5150 at 1.6ghz. The Core 2 was almost twice as fast in the single threaded test, but they were about equal in multithreaded. Granted Cinebench is not a very good test, but the results are pretty much what I expected: the core 2 is faster in anything except workloads that can perfectly use all 4 cores of the athlon, in which they are about equal.

 

[Turbonium]

Interesting.

Anyone else who wants to pitch in, please do.

 

[AnonymouseUser]

The biggest difference is the supported instruction sets (including 64-bit extensions) and the iGPU. I would also suspect the A4 to be a much better multi-tasker.

Comparison

 

[Turbonium]

All other things being equal (power draw, heat/noise, etc.), and only taking into consideration performance, which of the two CPUs would you rather for "heavy" general use in a Windows environment (no gaming)?

I'm not sure what raw benchmark would help me answer this. I'm assuming it's a combination of both single- and multi-threaded performance. Also, I don't think instruction sets have any huge bearing on this sort of comparison.

I mean, for Windows, would you rather 2 superior cores? Or 4 significantly lesser cores?

 

[Jimzz]

I would take the AMD in this compression. As said more updated instructions and better iGPU as well that will remove some of the work from the CPU.
After that add a SSD and it should be a very efficient system.

 

[frozentundra123456]

Barring some use that specifically utilized the more modern instruction sets of the A4-5000, and if there was some kind of decent gpu with it, I would go with the Core 2 Duo. We use similar desktop cpus in our lab computers with discrete low end gpus, and they are adequate for general use. If you were reduced to motherboard graphics, that might change.

I mean, at best, the A4 will be equal in perfectly threaded workloads, while the Core 2 will be faster in lightly threaded loads. I guess bottom line, is I just cant bring myself to accept these netbook processors (kabini and atom) in desktops.

 

[sm625]

They are not at all the same. They do not have very similar passmark scores. A4-5000 has a passmark single thread score of 584, whereas the T9500 has a score of 1063. That is way faster. Single thread performance is 90% of real world cpu performance. Multithread performance doesnt because you generally arent waiting on multithread performance. Multithreaded tasks are things you can get up and walk away from, like video encoding. It doesnt matter if it takes 2 minutes vs 3 minutes, or 2 hours vs 3 hours because you arent going to be sitting there waiting for it either way. But whether a page loads in 200 milliseconds or 400 milliseconds, you are going to feel that.

 

[TeknoBug]

I'm using an A4 5000 laptop, it'll have more instructions than the C2D, but that's about it, it's not bad for word processing and surfing the web and the occasional games (like Asetto Corsa and Guild Wars 2). Other than that, it's sluggish on loading things (although slapping in a SSD elimiated that).

Having tested out a Celeron N2930, it's somewhat identical to the A4 5000 and is a little cheaper, so have at it with your brand preference.

 

[Abwx]

Multithreaded tasks are things you can get up and walk away from, like video encoding. It doesnt matter if it takes 2 minutes vs 3 minutes, or 2 hours vs 3 hours because you arent going to be sitting there waiting for it either way. But whether a page loads in 200 milliseconds or 400 milliseconds, you are going to feel that.

When i open a page with Firefox on W8.1 the four cores are kicking, usage of each core is low but they are all used.

 

[Turbonium]

I guess bottom line, is I just cant bring myself to accept these netbook processors (kabini and atom) in desktops.

I used a C-60 for a long while (Ontario). Really slow. Kabini is obviously significantly more powerful, but it's more or less the same product tier. Based on that, and my current experience with the T9500, I felt the Core 2 Duo should be noticeably faster in real life. The benchmarks seemed to disagree though.

 

[NTMBK]

I used a C-60 for a long while (Ontario). Really slow. Kabini is obviously significantly more powerful, but it's more or less the same product tier. Based on that, and ny current experience with the T9500, I felt the Core 2 Duo should be noticeably faster in real life. The benchmarks seemed to disagree though.

It's double the cores, 50% higher clock speed, and 30% better IPC. It's like comparing Nehalem to a dual core Conroe- yes, it's the same tier, but it's a hell of a lot better.

 

[sm625]

When i open a page with Firefox on W8.1 the four cores are kicking, usage of each core is low but they are all used.

And? I have a 4 core 8 thread 2.8GHz cpu here with triple channel memory, and a meager 2 core 2 thread 4.5GHz haswell pentium cpu at home. The haswell pentium absolutely smokes this lug. It is day and night. It is literally 5 times as fast. I can launch 4 browsers, each with 10 tabs and 5 other apps twice as fast on the G3258 before this thing can even load a firefox with 10 tabs.

 

[Abwx]

And? I have a 4 core 8 thread 2.8GHz cpu here with triple channel memory, and a meager 2 core 2 thread 4.5GHz haswell pentium cpu at home. The haswell pentium absolutely smokes this lug. It is day and night. It is literally 5 times as fast. I can launch 4 browsers, each with 10 tabs and 5 other apps twice as fast on the G3258 before this thing can even load a firefox with 10 tabs.

What is the relevance of your 4.5GHz CPU..?

I could tell you as well that a 2C 4Ghz is probably faster than a 4C 2Ghz.

That said i have an Athlon 5350, granted it is 2GHz, and a Pentium 2C T4400 2.2, the latter is sluggish compared to the former.

 

[Turbonium]

It's double the cores, 50% higher clock speed, and 30% better IPC. It's like comparing Nehalem to a dual core Conroe- yes, it's the same tier, but it's a hell of a lot better.

I'm not disagreeing, that's why I looked up benchmarks, then asked here. I wasn't expecting Kabini to be that good, at least on paper.

 

[SPBHM]

as a CPU I would go with the much higher ST performance from the t9500, but as a platform the a4 5000...
the IGP have proper video acceleration and good UI performance (I think), t9500 probably have the HD 4500... so it's not an easy choice, also if both have comparable battery, screen and so on, battery life could be a big win for the A4.

 

[frozentundra123456]

You are probably correct. I still think the T9500 would be faster as a pure cpu, everything being equal, which is technically what the OP was asking. But if he was really asking if a modern platform with the A4 would be more responsive than an older laptop with the T9500, it is quite possible that it would, due to better thermals, igpu, ram, and storage.

 

[Turbonium]

as a CPU I would go with the much higher ST performance from the t9500, but as a platform the a4 5000...
the IGP have proper video acceleration and good UI performance (I think), t9500 probably have the HD 4500... so it's not an easy choice, also if both have comparable battery, screen and so on, battery life could be a big win for the A4.

In my case, it's the X3100 (GM965).

You are probably correct. I still think the T9500 would be faster as a pure cpu, everything being equal, which is technically what the OP was asking. But if he was really asking if a modern platform with the A4 would be more responsive than an older laptop with the T9500, it is quite possible that it would, due to better thermals, igpu, ram, and storage.

Yea, I was asking only in terms of the CPU, because I already have this T9500 system. All I need to do is max out the RAM to 4GB ($70), and throw in a new battery ($150). So $250 at most (or I could just keep it at 2GB RAM and not go over $200).

Going the A4-5000 route (with 4GB of RAM) means getting a whole new system for $650+. Even if the platform is better, it's not really worth it if I already have a working platform that does the job, especially if this T9500 (and thus the entire system) is actually noticeably faster in many situations.

 

[frozentundra123456]

650.00 for an A4 system seems outrageously expensive. I recently bought a new laptop for my wife with a standard voltage mobile i3 (2.4ghz I think) for 360.00.

 

[NTMBK]

Don't forget that Kabini integrates 6GB/s SATA, which will make for a more responsive SSD- that will speed up your day to day experience.

 

[sm625]

Don't forget that Kabini integrates 6GB/s SATA, which will make for a more responsive SSD- that will speed up your day to day experience.

Is there any benchmarks showing that SATA on the SoC has lower latency or higher transfer rates?

 

[Abwx]

Is there any benchmarks showing that SATA on the SoC has lower latency or higher transfer rates?

http://www.guru3d.com/articles-pages/amd-athlon-5350-apu-and-am1-platform-review,15.html

 

[pw257008]

What is the relevance of your 4.5GHz CPU..?

I could tell you as well that a 2C 4Ghz is probably faster than a 4C 2Ghz.

That said i have an Athlon 5350, granted it is 2GHz, and a Pentium 2C T4400 2.2, the latter is sluggish compared to the former.

Yeah, my brother had a Core 2 laptop Pentium and it was rough. We could never figure out exactly what the problem was, but maybe the cache? Of course I'm sure bloatware didn't help matters.

 

[Abwx]

Yeah, my brother had a Core 2 laptop Pentium and it was rough. We could never figure out exactly what the problem was, but maybe the cache? Of course I'm sure bloatware didn't help matters.

Actualy i never checked the SATA transfert numbers of the AM1 plateform, from the graphs above i think that the responsivity has a lot to do with HDD/SSD bandwith, in this respect the old laptops like ours will be handicapped due to the older HDD/SSD port bandwith, apparently the AM1 board manage to get 4 Gbit/s speed out of the theorical 6Gbit/s.