Oh how I love me some Core2Duo...

[VirtualLarry]

Yes, I know that the architecture is like eight years old by now. But it has really stood the test of time.

I have a Netbook with an AMD C-60 in it, and a HTPC with a C-70 in it, and they are just Waaaay too slow for my liking. Newest version of Skype now uses 100% of CPU time on the C-70, and the person on the other end reports that my voice breaks up. Previous versions of Skype ran at 85% CPU time, and at least the convo was clear.

I haven't tried Skype on it, but I recently unboxed an HP / Compaq DX7400 refurb, with an E6550 2.33Ghz C2D with 4MB L2, and just web browsing, it was sooo smooth, compared to the C-70. Even with half the RAM, and a HDD rather than an SSD.

I really missed my Core2 era CPUs, compared to all of these power-saving backwards-performance-looking crap chips, like the C-60 and C-70, and I suppose the 1.0Ghz dual-core Kabini too. (Atoms too, although I haven't really had a chance to try a new 22nm OoO Atom dual-core yet.)

Even my Ivy Bridge Celeron 1007U 1.5Ghz dual-core that I'm typing this on (with SSD), feels slightly slower than that 2.33Ghz 65nm C2D CPU.

So long-live C2D, and I hope that chips get faster, and not slower.

 

[Soulkeeper]

I share your sentiment.
I own a c-50 laptop, I run it as a lightweight webserver because it only pulls like 9 watts with an ssd in it.

Compared to my old Q9450 I'm not convinced that anything AMD has to offer outside of fm2/fm2+ _definitively_ beats it. I mentioned this months back in a forum thread and they said I needed "medication for that" ... ohh well

 

[NTMBK]

I share your sentiment.
I own a c-50 laptop, I run it as a lightweight webserver because it only pulls like 9 watts with an ssd in it.

Compared to my old Q9450 I'm not convinced that anything AMD has to offer outside of fm3/fm3+ _definitively_ beats it. I mentioned this months back in a forum thread and they said I needed "medication for that" ... ohh well

Considering that even Trinity beats that CPU on almost every benchmark, I'm not surprised: http://www.anandtech.com/bench/product/51?vs=675

 

[AtenRa]

No need for Core 2 Duo in 2014, 2GHz Quad Core AMD Kabini or Intel 22nm 2GHz ATOMs are fine in every day PC usage like browsing etc.

C-60 and C-70 were the lowest power APUs of the Bobcat family, they were fine in 2011 but not anymore.

 

[VirtualLarry]

No need for Core 2 Duo in 2014

If only... AMD could design a CPU with IPC and ST performance as good as a 45nm Core2 CPU... then that might be true.

2GHz Quad Core AMD Kabini or Intel 22nm 2GHz ATOMs are fine in every day PC usage like browsing etc.

I wish!

C-60 and C-70 were the lowest power APUs of the Bobcat family, they were fine in 2011 but not anymore.

They were abysmal, performance-wise, when they were first released.

Looking forward to a 3.0Ghz Puma-core quad-core APU on AM1.

Don't get me wrong, I'm going to probably get a 5350 and an ASrock AM1 board just for kicks. But I doubt it will match the performance of my E3300 2.5Ghz Core2 CPUs.

I wonder too, how C2D and Bobcat have aged, in the 64-bit OS era. I wonder whether they were optimized for 32-bit or 64-bit execution.

Edit: The reason that I bring up that last point, is that the DX7400 with the E6550 2.33Ghz Core2, was running 32-bit Win7, and the C-60 and C-70 are running 64-bit Win7.

As everyone knows, 64-bit data and pointers take up twice the size in cache memory, making it half as effective. Plus the E6550 has 4MB L2, and the C-70 has 1MB per core (non-shared, AFAIK).

 

[Soulkeeper]

Considering that even Trinity beats that CPU on almost every benchmark, I'm not surprised: http://www.anandtech.com/bench/product/51?vs=675

2.66GHz
most of us here who bought a core2q would be ashamed to run it under 3GHz

software not specifically compiled with avx fma support can't unlock the "breakeven" potential of these APUs comapred to core2q clock for clock.
http://www.anandtech.com/show/7711/...f-kaveri-and-other-recent-amd-and-intel-chips

 

[Matt1970]

C-60 and C-70 were the lowest power APUs of the Bobcat family, they were fine in 2011 but not anymore.

The C-60 and C-70 were never fine.

 

[NTMBK]

2.66GHz
most of us here who bought a core2q would be ashamed to run it under 3GHz

software not specifically compiled with avx fma support can't unlock the "breakeven" potential of these APUs comapred to core2q clock for clock.
http://www.anandtech.com/show/7711/...f-kaveri-and-other-recent-amd-and-intel-chips

And Trinity can overclock too. :hmm: As for FMA- peak floating point throughput is the limiting factor in a few very specific workloads. There are plenty of other factors like integrated memory controllers, branch predictors, caches, etc etc. The benches don't lie.

 

[NTMBK]

I wonder too, how C2D and Bobcat have aged, in the 64-bit OS era. I wonder whether they were optimized for 32-bit or 64-bit execution.

Edit: The reason that I bring up that last point, is that the DX7400 with the E6550 2.33Ghz Core2, was running 32-bit Win7, and the C-60 and C-70 are running 64-bit Win7.

As everyone knows, 64-bit data and pointers take up twice the size in cache memory, making it half as effective. Plus the E6550 has 4MB L2, and the C-70 has 1MB per core (non-shared, AFAIK).

I suspect that SSE performance will play a big part, too. Bobcat only had 64-bit vector units, which severely limited its SSE performance, and on 64-bit you are guaranteed a minimum of SSE2 support which means that 64-bit software is much more likely to use SSE1/2 instructions than 32-bit.

 

[OBLAMA2009]

i loved the core 2 era. it was the last time cpu's were sold by clockspeed. now theyve released all sorts of slower chips that you dont know how to compare, and they dont even advertise clock speeds, so ive just lost interest

 

[Soulkeeper]

The problem with benchmarks is you can always look at the one which justifies your stance in an argument that doesn't need to exist.
CPU floating-point peak performance is mainly what i'm concerned with, but that's just me. Two people are unlikely to look at or care about the same benchmarks.
You can't exceed the limits of what your x86 extensions offer, kinda like a maximum speed limit. Most software won't hit that peak, but it'll never exceed it.

It's not hard to write optimized software which runs on a 4 issue 4 core cpu with 12MB of cache faster than on a 2 module APU with 4MB of cache.
I'm just trying to take an academic view of the designs here.

 

[strata8]

Very silly to buy a netbook with a 40nm 9W chip and then complain that it doesn't match the performance of your 65nm 65W chip from a few generations back.

I mean, holy hell. An i3 3220 will absolutely destroy that E6550 and use less power to boot. From the tone of your post you'd think that AMD and Intel have discontinued their higher end chips and are forcing people into netbook-class CPUs.

 

[VirtualLarry]

From the tone of your post you'd think that AMD and Intel have discontinued their higher end chips and are forcing people into netbook-class CPUs.

No, but there is an obscene proliferation of those "netbook chips" on the market. Many of those AIO PCs have them. I wonder if someone took a weighted average of PCs sold, and the CPUs contained within, whether the overall performance average is decreasing over time.

Would that be a sign that Moore's law (the laymen's version) is going backwards?

 

[NTMBK]

The problem with benchmarks is you can always look at the one which justifies your stance in an argument that doesn't need to exist.
CPU floating-point peak performance is mainly what i'm concerned with, but that's just me. Two people are unlikely to look at or care about the same benchmarks.
You can't exceed the limits of what your x86 extensions offer, kinda like a maximum speed limit. Most software won't hit that peak, but it'll never exceed it.

It's not hard to write optimized software which runs on a 4 issue 4 core cpu with 12MB of cache faster than on a 2 module APU with 4MB of cache.
I'm just trying to take an academic view of the designs here.

Oh yeah, if you're writing scientific vector-FP heavy code I can easily see the C2Q winning unless you break out the FMADDs. The Bulldozer family just aren't that great at it. Rumour has it that Excavator will double the vector widths and have proper 256-bit units, but we'll have to wait and see.

 

[Keysplayr]

Inspiron 1545 with 2GHz Core2Duo here still going very strong. Fast, snappy. Running Windows 7 64. 3GB DDR2 SODIMMS. 128GB Crucial SSD.

Powerful enough for recording in ProTools, Cubase and Sonar (I use all 3).
I like me some C2D too.

 

[NTMBK]

No, but there is an obscene proliferation of those "netbook chips" on the market. Many of those AIO PCs have them. I wonder if someone took a weighted average of PCs sold, and the CPUs contained within, whether the overall performance average is decreasing over time.

Would that be a sign that Moore's law (the laymen's version) is going backwards?

Unfortunately that's the usual PC OEM mentality of "race to the bottom", "maximise margins", while destroying the PC market long term because they are associated with godawful performance and terrible value. Anyone who got a Bobcat-powered desktop is liable to be put off PCs for life, when they can get an Android tablet which is appreciably faster.

You can sometimes find good models- I helped my parents buy a Packard Bell compact PC (half height PCIe slots, mini-ITX motherboard) which crammed a Sandy Bridge i5 into it at an affordable price. But there are plenty of stinkers out there.

 

[dave_the_nerd]

i haz Macbook. 2.13GHz C2D.

Is Happy.

 

[Yuriman]

45nm C2Q generally falls somewhere just below Piledriver, per clock. 65nm chips were good for 3-3.6GHz, while 45nm chips were often good up to 4GHz or beyond with good cooling - fast enough to be "fast enough" even today. There's no doubt AMD's current offerings are both faster and more power efficient, but a 2 module chip would not be much of an upgrade.

 

[Insert_Nickname]

It's not hard to write optimized software which runs on a 4 issue 4 core cpu with 12MB of cache faster than on a 2 module APU with 4MB of cache.
I'm just trying to take an academic view of the designs here.

Yorkfield's massive L2 certainly doesn't hurt anything. Most never duals only have 3-4MB, even less for Celeron/Pentiums. Sandy/Ivy/Haswell quads "only" have 256KB L2, with 8MB L3.

As a side-note I'm really impressed just how well the (Dothan) Pentium M has aged. Perhaps that's partly because of the massive (for the time) 2MB L2.

Any thoughts on this matter...?

 

[Maximilian]

Yeah Core2 was awesome. I remember the E6xxx's coming out and demolishing the athlon X2's and people wanting the E6600 because of its 2MB extra cache but others arguing the E6400 is just as good and it didnt matter etc etc, good times

Got an SU2300 celeron in my laptop and its pretty average.

http://ark.intel.com/products/42779/Intel-Celeron-Processor-SU2300-1M-Cache-1_20-GHz-800-MHz-FSB

 

[2is]

Very silly to buy a netbook with a 40nm 9W chip and then complain that it doesn't match the performance of your 65nm 65W chip from a few generations back.

I mean, holy hell. An i3 3220 will absolutely destroy that E6550 and use less power to boot. From the tone of your post you'd think that AMD and Intel have discontinued their higher end chips and are forcing people into netbook-class CPUs.

I agree. OP makes it seem like there is no mobile CPU solution to match the powererformance ratio of a C2D which couldn't be further from the truth. Yes, there are newer chips that are slower than C2D, but it's your decision to buy that when you walked into the store and decided you wanted the clearance item laptop priced at $300.

While I agree that C2D's were great, and I'm still sporting one in my 2010 MacBook Air, they're easily outclassed in every metric imaginable with todays offerings.

 

[Arkaign]

Conroe and all it's forms is an absolutely legendary processor line, for good reason (in the right configuration/clocks it can still do basically anything to this very day in a reasonable manner).

K7 and K8 were the same, epic products of their times, and viable even after their prime.

We're on the tail end now hopefully (for the sake of progress). The ultra-low performance processors are a bit weaker than I think deserve to exist outside of tablets. Something in the raw performance range of a 2.2Ghz C2D 2MB CPU should be the absolute floor in 2014, yet there are products from both Intel and AMD in the supply chain that can make a grown man cry from the sheer pathetic performance they yield.

I think 14nm will finally nail that coffin shut, there will no longer be any excuse for something that performs worse than a processor nearly a decade down the line, regardless of power consumption.

 

[escrow4]

Conroe and all it's forms is an absolutely legendary processor line, for good reason (in the right configuration/clocks it can still do basically anything to this very day in a reasonable manner).

K7 and K8 were the same, epic products of their times, and viable even after their prime.

We're on the tail end now hopefully (for the sake of progress). The ultra-low performance processors are a bit weaker than I think deserve to exist outside of tablets. Something in the raw performance range of a 2.2Ghz C2D 2MB CPU should be the absolute floor in 2014, yet there are products from both Intel and AMD in the supply chain that can make a grown man cry from the sheer pathetic performance they yield.

I think 14nm will finally nail that coffin shut, there will no longer be any excuse for something that performs worse than a processor nearly a decade down the line, regardless of power consumption.

Actually I'd say the minimum should be an E8400, or the direct modern descendant, the Haswell G1820.

 

[Vesku]

I'd still be using my Dell Core 2 notebook if it wasn't such a heat monster and the graphics were just a little less terrible (also hot when you actually make it do some light 3D). Although even if Dell had put excellent cooling in it there is still the issue that you'd need a really huge battery to get more than 2 or 3 hours of active usage out of it due to the power draw. Hmm, typing it out, only 1 out of 3 main notebook traits was acceptable with Core 2. It's amazing actually how it's only in the last 2 or 3 years that notebooks have actually become somewhat acceptable as an actual portable device.

Looking forward to Kabini's and BayTrail's successors, should be great for sub 15 inch travel notebooks. They should have better IPC than Core 2 with a big jump in graphics and at much less power usage.

Sandybridge and Ivybridge still a bit too power hungry for travel notebooks imo, i.e. not your primary or work machine. Haswell is good but pay a bit too much of a premium for my liking for it in sub 15 inch form.

 

[Arkaign]

Actually I'd say the minimum should be an E8400, or the direct modern descendant, the Haswell G1820.

Fair enough, and agreed on the same token.

I used something like this recently while setting up a user for Office 365, and it was absolutely astonishingly terrible :

http://www.walmart.com/ip/HP-Black-P...Opera/21634598

Things that should have taken at most 5-10 seconds took well over a minute. Opening three tabs + syncing a small 3GB outlook OST brought the thing to a nearly frozen state. I wanted to take a sledgehammer to it. And it cost nearly 500 bucks. The beater laptop I use for work (that I'm typing this on) is a lowly T4200 2Ghz C2D (Pentium dual-core branded), with 4GB DDR2, and it's like a lightning bolt by comparison in terms of usability. At present I have outlook 2013, a half dozen PDF tabs (Foxit), another couple dozen web tabs (Firefox and Chrome), and it's still instantly responsive to what I tell it to do.