LinkBack URL
About LinkBacksThat's what I think will happen - they'll move/improve process before changing the arch.Originally Posted by Corky34
That's what I think will happen - they'll move/improve process before changing the arch.
Why not? AMD have plenty of power efficient options out there already. If the new process allows for higher power parts then they might as well put that soldered IHS to good use. As long as it is cool when idle I don't see any problem. For most people efficiency at load only needs to be reasonable - i.e. the added expense and noise is buying a reasonable boost in performance.
I would like to see two improvements, though. Firstly a selectable TDP would be nice - I would happily run my R7 at 45W for general computing and 180W when performance matters. Secondly configurable boost clocks - since you have to be pretty brutal with an R7 or a weak R5 to get an all-core overclock that doesn't leave single-threaded performance on the table (by having the all-core clock at least match the max XFR frequency).
Last edited by CAPTAIN_ALLCAPS; 06-10-2017 at 10:13 AM. Reason: Correction
CAPS LOCK IS NOT A BUTTON IT IS A WAY OF LIFE.
If you have the money is it R5-1600X or i7-8700K?
Hm, terminology confusionThat's what I think will happen - they'll move/improve process before changing the arch.By Ryzen 2 I meant second generation Ryzen processors: i.e. Ryzen 2 will use the Zen+ core (i.e. minor tweaks + shift to 12LP process). Third generation Ryzen processors will use Zen 2.
I suspect there are some minor quick wins they can ring out of the architecture even from fairly minor shifts, and AMD have said from day one of the Ryzen launch that they know which areas of the arch they can optimise. I rather suspect that the rumoured double-clock on the infinity fabric is something they couldn't enable with the 14LPP process but might be possible on a different node, and that should drop the inter-CCX latency significantly...
kalniel (06-10-2017)
If it was my money i would go with a R5 1600 (non-x) and OC it, that way i could upgrade the CPU if i wished until 2020 without having to buy a new board.
With Coffee Lake you're basically looking at using a modified 3 year old chipset until Intel release the real 300 series in Q2 next year.
finally a price drop i can understand. however i am still a die hard AMD fan. its still does what i need it to do while i multitask in daily activity.
The situation with the motherboards has ruined any value comparison for intel. The 8400 might have the same RRP for the chip as the 1500X, but in a complete system there's easily a £70 difference. The actual competition for the 8400 is the 1600X, which is still £20 cheaper - and for just £40 more you can have a 1700. That £40 difference is using the stock intel heatsink, of course - if you want something quieter it'll eat into the cost difference.
Comparing the 1600x to the 8400, the only benefit is lower power consumption. In multi-threaded apps the ryzen is well ahead (no surprises there), and in gaming there's no clear winner - some benchmarks have intel ahead, some have AMD. Is 10w lower power at idle and 10-15w lower consumption under load really worth £20*?
* worth it after 19 months, if you leave the pc idling 24/7 (assuming 14p/kwhr), but who does that?
Motherboard and CPU for the 8700k costs literally double the price of the equivalent combo for the 1600x, they aren't comparable. An 1800x system is £30 cheaper than the 8700k system.
I wonder if the Coffee Lake i3 is just a rebranded Kaby Lake die, and therefore pretty much a KBL i5?
The 8100 has the same UHD IGP, so it can't be a rebranded die (unless they've been sitting on the silicon for openGL 4.5 for a while without turning it on). In practical terms it is just a rebranded i7 without turbo boost, but I don't think it's the same die
You mean i5? It doesn't have the SMT of the i7, which is more or less the only difference between them.
From what I've read, even the 'UHD' is basically a rebrand given the graphics part is fundamentally the same as its 'HD' predecessor. However, there may be a couple of hardware changes similar to what we saw in going from SKL to KBL e.g. some differences in hardware video decoding.
In terms of execution hardware, it does pretty much seem like a re-rehashed SKL i5, which isn't necessarily a bad thing for i3 money - I was just wondering whether they'd actually released something akin to a new stepping for it, or literally just stuck a new badge on it.
Edit: I've just remembered, the mobile KBL refresh parts also call their GPUs 'UHD': https://ark.intel.com/products/122589
You're right - I was getting confused with the mobile i5's. Confusing things further, the mobile UHD IGP's don't support openGL 4.5 like the desktop chips, only 4.4, so the extra U is just marketing.
The 8100 is a great chip for i3 money, if it were available for i3 money - as I've banged on about before, the motherboard situation puts it up against the 1600 or a last-gen i5 (7400 is £10 cheaper with b250, 7500 is £10 more expensive)
Yeah the motherboard situation makes it a lot less attractive. Lets just say for a moment Intel really needed to make platform/socket changes for the 6 core parts - what's the excuse for the quads?
Same socket for an entire family of chips? I mean seriously, can you imagine the reaction if Intel released a family of processors where the i3 needed a different family of motherboards to the i5 and i7?! I mean, there'd be an inevitable enthusiast backlash of the confusing nomenclature, and how the i3s are actually any different to the previous gen i5 if they use the same mobos and chipset ... and then you'd have the OEMs, who'd need to manage two different motherboard designs for this year's processors despite them theoretically being the same family of processors....
No, I think having the entire range on one socket makes sense. What doesn't make sense is not having any other chipsets ready to go at release. It rather smells like another rushed/mis-managed launch from Intel .. their second this year. Wonder what's got into them...
True, I didn't think that one through, did I?
However, having had a cursory glance at the pinouts, are they not just assigning additional pins that were previously unused, to power planes? If so, it should be theoretically possible to use them interchangeably. But that would mean admitting that adding a few more power pins doesn't really justify mandating a new platform in the first place - and one with the same name, just to confuse people.
Phoronix have run some benchmarks on the i3 8100. Looking at the scores, the i3 8350K should be quite close in performance to the FX 8350 but for about £40 more
No idea, I've not had a look at the pin out
Of course, AMD didn't claim you couldn't slap an AM3+ socket onto an older chipset, so we very quickly got a lot of cheap AM3+ boards using older AM3 and AM2 chipsets, often with the boards branded to make them sound like newer chipsets (i.e. all the "960" chipset motherboards which actually used the 760G chipset).
As I said earlier, the whole thing just smells somewhat of a rushed launch. I mean, the rehashed socket with no cross-compatibility isn't new to Intel - they did the same with s2011/v3 after all - but to launch with a single rebranded chipset is ... odd. It's hard to avoid the thought that Intel have pushed this out of the door early because Ryzen is cannibalising their mid-range sales....
There are currently 1 users browsing this thread. (0 members and 1 guests)
Posting Permissions
Reply With Quote