• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

Intel Core Ultra 9 285k 'Arrow Lake' Discussion/News ("15th gen") on LGA-1851

Associate
Joined
6 Mar 2013
Posts
535
Pity it couldn't have a Bloomfield vibe. My i7 920 with triple channel DDR3 was an absolute beast on launch day, so ahead of it's time. Had great overclocking headroom also - mine ran at 3.8Ghz all of it's long life.
How does triple channel DDR3 compare to Dual Channel DDR4? (1333 Vs 3600) Just curious. (12gb Vs 16gb)
 
Soldato
Joined
1 Feb 2006
Posts
3,437
Personally don't find the explanation convincing as to removing hyperthreading - it seems like they've ring fenced the benefits to talk themselves into it being a good thing, maybe even an almost religious adherence to some mantra they've come up with, rather than actual delivering a real benefit.
Hyperthreading has been on borrowed time for ages, the overhead vs benefits is low, more so as the core count grows. This is before thinking about the security issues. Some workloads do get big gains but at a cost. I would prefer 20 real cores vs 16 real + 16 logical.
 
Man of Honour
Joined
13 Oct 2006
Posts
91,697
Hyperthreading has been on borrowed time for ages, the overhead vs benefits is low, more so as the core count grows. This is before thinking about the security issues. Some workloads do get big gains but at a cost. I would prefer 20 real cores vs 16 real + 16 logical.

A lot of the stuff I do sees big benefits from HT and sadly not much in the way of consumer entry level HEDT platforms like X79 any more where it, or higher core count CPUs might exist.
 

RSR

RSR

Soldato
Joined
17 Aug 2006
Posts
9,615
That Tech Talk video is pretty interesting as, as a concept, it makes a lot of sense but as normal we need some real-world testing.

HT is an interesting one, as I have been testing it on and off and a few games (2042 / COD etc...) and for the most part it has not made a huge difference. As an example, the COD benchmark runs better with HT off but Warzone runs better with it on.
 
Soldato
Joined
15 Oct 2019
Posts
11,908
Location
Uk
I think Intel needs to drop HT to get power under control and also boost clock speeds for SC workloads, MT will take a bit of a hit but then Intel have a big MT lead in many price points 7600X vs 13600k, 7700X vs 13700k, 13900k vs 7900X for example saw intel sitting on a solid MT lead so they can afford to lose some MT and then there is all the HT exploit fixes that slow performance so Intel should pick up some performance here as well.
 
Man of Honour
Joined
22 Jun 2006
Posts
12,228
How does triple channel DDR3 compare to Dual Channel DDR4? (1333 Vs 3600) Just curious. (12gb Vs 16gb)
DDR3 1333 is 10.67 GB/s per channel.
DDR4 3600 s 25.6 GB/s per channel.

Calculate memory bandwidth:
Memory clock x channel (64-bit) = total bits per second
total bits per second / 8 = total bytes per second
total bytes per second / 1000 = total gigabytes per second
 
Associate
Joined
6 Mar 2013
Posts
535
So off the top of my head without calculating anything they are about equal? Or is the voddie clouding my math skills. (I 'was' top set in high school)
 
Last edited:
Associate
Joined
6 Mar 2013
Posts
535
DDR3 1333 is 10.67 GB/s per channel.
DDR4 3600 s 25.6 GB/s per channel.

Calculate memory bandwidth:
Memory clock x channel (64-bit) = total bits per second
total bits per second / 8 = total bytes per second
total bytes per second / 1000 = total gigabytes per second
Also I appreciate you giving us the actual statistics, but we're not all boffins. A simple explanation would suffice.
 
Soldato
Joined
3 Oct 2013
Posts
3,671
DDR3 1333 is 10.67 GB/s per channel.
DDR4 3600 s 25.6 GB/s per channel.

Calculate memory bandwidth:
Memory clock x channel (64-bit) = total bits per second
total bits per second / 8 = total bytes per second
total bytes per second / 1000 = total gigabytes per second

That calc looks wrong to me ,

Gigabyte is 1024 megabytes. Wouldn't you need to divide it 3 times to get gigabyte?
 
Last edited:
Associate
Joined
28 Sep 2018
Posts
2,301
Alder Lake (12th Gen) to Raptor Lake (13th Gen) did have a small performance bump, as the main gains came from the higher core clocks and the increased cache. The IO area also increased performance as it brought up the base speed on the RAM side plus overall 13th / 14th Gen have far better IMC's. Raptor Lake to Raptor Lake-R was more of a node refinement.

I seem to be able to push my 14900KS more than my 13900KS but that's probably just a refinement feature (Better Silicon). If you had an Alder Lake CPU and wanted to upgrade outside of the 14900K/KS the 14700K makes the most sense.

The only reason the refresh exists is because meteor Lake could not be clocked high enough to beat 13th GEN
 
Soldato
Joined
3 Oct 2013
Posts
3,671
Yes / 1024 = kilobytes, then / 1024 = megabytes then again by 1024 for gigabytes.
I was thinking I forgot how to divide when I read it :cry: Nice to know I've not gone nuts just yet:D

Edit:
That's not @humbug s alt account is it with them maths skills :cry:

Edit 2:

Realised I put a gigabyte as 1024 kilo instead of mega:o
 
Last edited:
Associate
Joined
30 Jan 2003
Posts
1,517
Location
Cardiff
13'th gen is nice and as good as that architecture got, which is to say it is good. Its the one to get if you're going Intel.

14'th gen is just desperate and should be avoided.

but since 14th gen is the same price as 13th, you might as well get 14th gen. Also there are some benefits with Intel's APO.
 
Back
Top Bottom