Is the i7-970 still a good choice?

[Burpo]

That performance test only shows a stock 3770, a stock 4790K no overclocking would be even faster and dump way less heat and power. Not even mentioning a 5820K.

lol.. at 5Ghz it was throttling thru the whole test! Cinebench results for X5650 @ 4.4Ghz is about the same as a stock 5820.
Note that X5650 is a 2.66Ghz chip, but when overclocked it keeps up with the 5820..

From our own Cinebench thread here..
http://forums.anandtech.com/showthread.php?t=2361017

"6 cores

10. biostud |i7-5820K | 4.4GHz | 13.68

16. Burpo | Xeon X5650 | 4.87GHz | 11.81

4 cores

1. crashtech | i7-4790K| 4.9GHz| 10.77"

i7-5820 stock scores 10.67.. An i7-970 @ 4.4Ghz scores 11.21, but for $60 the Xeon choice makes more sense for the OP

 

[Lepton87]

lol.. at 5Ghz it was throttling thru the whole test! Cinebench results for X5650 @ 4.4Ghz is about the same as a stock 5820.
Note that X5650 is a 2.66Ghz chip, but when overclocked it keeps up with the 5820..

From our own Cinebench thread here..
http://forums.anandtech.com/showthread.php?t=2361017

"6 cores

10. biostud |i7-5820K | 4.4GHz | 13.68

16. Burpo | Xeon X5650 | 4.87GHz | 11.81

4 cores

1. crashtech | i7-4790K| 4.9GHz| 10.77"

i7-5820 stock scores 10.67.. An i7-970 @ 4.4Ghz scores 11.21, but for $60 the Xeon choice makes more sense for the OP

What? 10.67? HOW? Are we talking about 11.5 you linked? I score 11.63 on it.

 

[Burpo]

Cool.. Good to know.. Thank you.. I had only seen one video of a cinebench run on a stock 5820K and it scored 10.67..

45 seconds into this video..
https://www.youtube.com/watch?v=zi3r5wMjGm4

 

[MongGrel]

*lurks* .....

 

[Lepton87]

Cool.. Good to know.. Thank you.. I had only seen one video of a cinebench run on a stock 5820K and it scored 10.67..

45 seconds into this video..
https://www.youtube.com/watch?v=zi3r5wMjGm4

strange, that score was with the RAM set at 2133 instead of 2667MHz, I'll see if it makes any difference. MOBO might make a difference, ROGs are one of the better ones.



11.78 seems I matched your score at stock but something is wrong with my RAM it crashed after a test at 2600MHz this score is with RAM at 2400MHz, I bought it as 2667 but SPD only shows 2133

 

[Burpo]

Your Asus board is likely the difference. Surely you see that the overclocked $60 Xeon has similar performance as the stock 5820k now..

Just ran it for the 1st time on Windows 10.. Not the 40% improvement you had hoped for?

 

[YBS1]

Sure, and it's stable at 4.85GHz? It was basically a suicide run, right? I don't have time to OC right now, but I'll see how much I can manage but it's just one benchmark, if we used linpack with AVX2 then... You know what I mean.

http://anandtech.com/bench/product/1320?vs=444

Sometimes HW advantage is huge sometimes very small, 11.5 are one of the better cases for Westmere. Why 11.5 and not 15 or 10? Easy to see with the benchmarks above.

This is my 24x7 setting on my 5670, 195x24=4690MHz. The suicide runs are listed in the benchmark threads. Probably about on par or a little faster than a typical stock clocked 5820K from a quick Google search (showed a range of about 10.7-11.2). The OP should probably note, both Burpo's chip and mine would be considered atypical, seems most Westmeres settle in the 4.2-4.4 range.

 

[Lepton87]

This is my 24x7 setting on my 5670, 195x24=4690MHz. The suicide runs are listed in the benchmark threads. Probably about on par or a little faster than a typical stock clocked 5820K from a quick Google search (showed a range of about 10.7-11.2). The OP should probably note, both Burpo's chip and mine would be considered atypical, seems most Westmeres settle in the 4.2-4.4 range.

That's still slower than my CPU at stock and as you said it's a very good chip. I get 11.78 at stock, quite a bit more than 10.7...

 

[Burpo]

Yeah, but how much is the difference worth? Not $300 difference..

Burpo's chip and mine would be considered atypical, seems most Westmeres settle in the 4.2-4.4 range.

FWIW, I don't believe it has as much to do with the chip as it does the motherboard supplying the chip with power..

At these speeds, we're throwing lots of current at these bits of silicon.. Board & power supply have to be up to snuff..

 

[Lepton87]

Yeah, but how much is the difference worth? Not $300 difference..



FWIW, I don't believe it has as much to do with the chip as it does the motherboard supplying the chip with power..

At these speeds, we're throwing lots of current at these bits of silicon.. Board & power supply have to be up to snuff..

Not that easy to put it in dollars some people are willing to pay 600$ more for 2 extra cores and it's totally worth it to them.

 

[YBS1]

That's still slower than my CPU at stock and as you said it's a very good chip. I get 11.78 at stock, quite a bit more than 10.7...

I'm going to take a guess at why you are scoring so much better than all the "stock" review scores from around the web. Not being familiar with your particular board but based on what I have seen from my own Asus X79 board enabling XMP profiles is probably allowing your chip to 6 core turbo all the time to what would normally be the chip's 1-2 core multiplier.

 

[BTRY B 529th FA BN]

Can I ask wtf you guys are calling stock? I've read alot of mis-info on here already

4.85GHz Westmere is a golden chip like 5.0GHz-4.7GHz Cache HW-E.

???

I get 11.78 at stock

??? where? and what does stock mean to you? i don't think this was out of context

 

[Lepton87]

Can I ask wtf you guys are calling stock? I've read alot of mis-info on here already



???

HW can clock its cache independently of its core speed like nehalem, it was taken to full core speed with Sandy and then with HW-E they reverted to a separate cache speed.

By stock I mean put the CPU in the cooler on and no fiddling with the BIOS.

I'm going to take a guess at why you are scoring so much better than all the "stock" review scores from around the web. Not being familiar with your particular board but based on what I have seen from my own Asus X79 board enabling XMP profiles is probably allowing your chip to 6 core turbo all the time to what would normally be the chip's 1-2 core multiplier.

Yeah, MCE was enabled but it is on most boards by default. Anyway my first OC attempt
4.4GHz 3GHz cache

 

[Burpo]

Not that easy to put it in dollars some people are willing to pay 600$ more for 2 extra cores and it's totally worth it to them.

It's totally worth it to me for $60.. :whiste:

Disappointed there's no 40% increase in 5 yrs?

Yeah, the way it's been going with IPC improvements between generations that have been on the order of 5-10% except sandy which was 15%, 35% higher IPC is a massive improvement. HW-E overclocks too and a bit higher so 40% faster is massive for a CPU, how long will we have to wait to get a CPU that is 40% faster than HW-E at 4.6GHz? (in single thread) 5 years? If those CPUs are similar than 8370(or 9570) is similar to 5930K in performance too.

Haswell E overclocks higher than a 2.66Ghz chip overclocked to 4.6Ghz? OK.. Show me the money!

 

[YBS1]

Well "stock" in the case of a 5820K would be boosting up to 3.4GHz under a 3-6 core load and up to 3.6GHz under a 1-2 core load. I'm guessing his board acts much like mine and with XMP enabled it just says "fug Intel specifications" and boosts to 3.6 even with a 6 core load. Nothing at all wrong with that, free performance for those who don't overclock.

 

[Lepton87]

It's totally worth it to me for $60.. :whiste:

Disappointed there's no 40% increase in 5 yrs?

There is between similarly OCed CPUs, that's still not much... You can't compare my tame 4.4GHz OC with a Westmere at 4.85, compare it to a westmere at 4.2GHz which would be more comparable and it should be compared across a wider variety of benchmarks.

 

[Lepton87]

Well "stock" in the case of a 5820K would be boosting up to 3.4GHz under a 3-6 core load and up to 3.6GHz under a 1-2 core load. I'm guessing his board acts much like mine and with XMP enabled it just says "fug Intel specifications" and boosts to 3.6 even with a 6 core load. Nothing at all wrong with that, free performance for those who don't overclock.

That would be under Intel spec, yes, but it just clocks to 3.6GHz under any load out of the box so its pretty much stock to me.

 

[Burpo]

Where does your i7-5820 score? Show me 16.5 (40% improvement)..

Epic FAIL!

 

[Lepton87]

It's benchmark specific, as anyone knowledgeable knows. But even in cinebench 11.5 HW-E at 5GHz would be close to 40% faster than your westmere, because I think such samples are comparable.




Here's what I managed so far

We should forget about large improvements in CPU performance especially in ST, single digits between generations, that's the sad truth unless you use new instructions than improvements are huge, let's not even try to compare Gflops in Linpack

 

[Burpo]

Never mind.. smh.. YOUR words were 40% and so far reality is 20%..

OP I hope you've seen the light.. don't waste your $60 on an upgrade for your system.. Get the latest & greatest and spend $1000+ so you too can be 20% faster..

 

[Lepton87]

Never mind.. smh.. YOUR words were 40% and so far reality is 20%..

OP I hope you've seen the light.. don't waste your $60 on an upgrade for your system.. Get the latest & greatest and spend $1000+ so you too can be 20% faster..

I didn't say in what benchmark and I had in mind a Westmere at 4.2GHz which is about average and I was right.

BTW. it's more than 20%, when we get 5-7% improvements between generations even 3% matters.

 

[Burpo]

Ok, do the MATH.. My rig is sitting right here next to me running Cinebench at 11.51 ( see above score). Your score is 14.44 (see above score). Are you ready now? This is the tricky part.. open calculator and type in 1151. Divide that by 1444.. What do you get? 79.7% .. meaning an improvement of 20.3%.. Still with me? DO THE MATH! Your claims of 40% are WRONG!

 

[Lepton87]

Ok, do the MATH.. My rig is sitting right here next to me running Cinebench at 11.51 ( see above score). Your score is 14.44 (see above score). Are you ready now? This is the tricky part.. open calculator and type in 1151. Divide that by 1444.. What do you get? 79.7% .. meaning an improvement of 20.3%.. Still with me? DO THE MATH! Your claims of 40% are WRONG!

I get 14.44/11.81 which is just under 23%. And I assumed a clock speed advantage for HW-E not Westmere, that what happens when such things are reversed.

Can't even do math, can you?

14.44/11.51 would be even more of an improvement, like 25%, you are calculating how much one cpu is slower than the other not the other way around.

 

[Burpo]

Whatever.. NOT the 40% you claimed now is it..

Note that X5650 is a 2.66Ghz chip, but when overclocked it keeps up with a stock 5820..

^^^^^ NOT WRONG! Put that in your Haswell koolaid & drink it..

 

[Lepton87]

Whatever.. NOT the 40% you claimed now is it..



^^^^^ NOT WRONG!

YEah, whatever 20% or 25% whatever lol good night, I had a good laugh, now just like you did math in reverse, compare the clock speeds in reverse as that is what you typically get from those CPUs hence my 40%