J7SC_Orion

...the caches are really, really helpful, but there are (rarer) limitations at the Azure MS FS 2020 server side. For example, if you fly around a really remote area with structures such as mountains that has very little air traffic, you'll notice a lot more 'tiles filling in' as you go compared to a new flight to a new region (for your system), but one which other folks seem to have crossed multiple times already. I'm speculating, but figure it must be Azure servers looking for less-often called-up bits from their multi-petabyte global map. I noticed this when I flew to the South Pole, as well as the Arctic...

...the same chap now posted 'Firestrike Extreme' today with his 6800XT...also 1st place at HWBot, but with a reduced margin over 3090 (both on LN2). I guess 'Firestrike Ultra' (4K) might be telling if/when he runs/posts that...

Yeah, 'rolling cache' (set to 200GB on my setup) is 'first-in, first-out' for storing scenery etc, and 200 GB is reasonably generous to hold quite a lot. There's also a 'manual cache' which is supposed to be more permanent, and with it you can pre-select areas on a map (with differing levels of details as an option) to load and store on your system locally. However, manual cache had some well-publicized issues early on - and using a big enough rolling cache should suffice / come out to almost the same thing - my 'local flights' are very swift due to the information stored in the rolling cache. Still, after various recent patches by MS for FS 2020, I should revisit the manual cache option as well. As to DX12 vs DX11; I do not expect the visual differences to be overly dramatic (given the superb graphics w/ DX11 already, at 4K/Ultra), but ray tracing could be interesting w/ DX12. Another benefit of DX12 is supposed to be the CPU cores utilization...currently with DX11, a 4C-8T is probably good enough as long as it clocks high, even better: a 6C physical cores high-clock CPU. But beyond that core count, not much to gain. However, I've read that with DX12, even 8C+ procs will have much better utilization and load distribution. ...and a bit of Arctic flying: Asgard mountain (also of James Bond fame) on Baffin Island

Tx for the update; no rush...I still have plenty of other containers to spill onto keyboards and mice ?

...noticed that a 6800 XT (w/ 5950X, both on LN2) took 1st place at Firestrike / Hwbot. While Firestrike is lower rez, still a big achievement, especially when considering the margin over 3090 (also LN2). HWBot is not office 'productivity software', but still...the basic design of RDNA2 seems to have a lot of headroom (at lower rez, at least).

...some brief flying in the snow around the neighborhood...after a while, wings etc really do ice up in MS FS 2020 ?

The NVidia ecosystem is the big one for AMD, IMO, re. productivity software. On the hardware side, I already mentioned potential memory bandwidth questions at 4K with the latest Radeon, which is otherwise very impressive (and priced right, comparatively speaking that is). I used a Radeon 79xx CrossFire system for years for both gaming and some 'light' productivity w/o any BSOD or other issues, and I know programmers who still swear by their Radeon Vega Frontier Edition on Ubuntu and custom video compression software. At the same time, the NVidia ecosystem is just far superior, at least for now, which is why most of our current systems are NVidia-based. It should be interesting to watch what happens with AMD Radeon on the 3rd-party software support side over the next few months, and also with NVidia Ampere 2.0 (TSMC 7nm) on the hardware side.

banned coz...who reads memos ? Possibly, programmers spent too much time in the real world tribute car, instead of coding ?!

I still have a single-phase cooler sitting in a corner a few yards from me...it used to settle in nicely at around -50 C for 4C/8T consumer CPUs, but with big HEDTs, its fundamental weakness (load response time) - mucked things up with the big CPUs...while it can easily handle 400+ W CPUs, it would take too long to react as loads shot up, by which time the system had crashed...always had to 'trick it' via pre-loading slowly to max power, then run a bench such as Cinebench

...agree that it would sell well. Apart from a memory bus bandwidth advantage, HBM2 uses less power than GDDR6(+X), but is more expensive..yet with an extra $300 to play with while undercutting the 3090 (for now...), it would have been my dream-card for 4K etc...a regular GDDR6 equipped-card could have been sold as is as the RX 6900, while the RX 6900 'XT' would have offered HBM2 (and perhaps Infinity Cache on top) But who knows, may be there will be a workstation / enterprise version coming out with HBM2...either way, a good time for upgrades in 2021 with this suddenly more competitive market.

Judging by 6800 XT test results so far, IMO 6090 XT performance will be very close to if not exceed 3090, at least at below 4K. I still wonder though why AMD didn't do a $1,299 6900 XT w/ HBM2 and its memory bandwidth to oppose the $1,500 3090

AMD 5959X all-core 16C/32T at 5.3GHz (GPUz validation) - with a helping hand from - Intel ?

...still more reading and reviewing to do, but it seems that at lower resolutions, the 6800XT is very strong while at 4K, memory bandwidth limitations start to come into play, even with Infinity Cache. Then again, I expect multiple driver updates, patches and the like...another reason to wait a bit as to how it all shakes out between AMD and NVidia EDIT: ...added Hardware Unboxed review (a lot of 1440p and 4K gaming data). As my top system is exclusively 4K, I am a little bit disappointed re. memory bandwidth. While the upcoming 6900 XT will be a real competitor to the NVidia 3090, I think it has the same memory setup as the 6800 XT...Infinity Cache clearly was deemed necessary for good reason, but I can't help wonder what a 6900 XT with HBM2 would do...

...another review of the 6800 XT below. Elsewhere, I think I read ('Google translate') that the RX 6K series do support CrossFire...leaving out the pros and cons of CrossFire, SLI etc, for a moment... a nice Ryzen 5950 build w/ 2x RX 6900 XTs and full w-cooling for all components might be an interesting project for early '21...but a lot more reviews of both AMD and NVidia to come before then...

I'm sure glad that I haven't run into any 4K limits whatsoever with my 2x factory w-cooled 2080 Tis - not because I don't like the new stuff / new builds, but because it affords me plenty of time (say, February '21) to pick and choose the best option w / o having to madly hit F5 for weeks on end only to have the pleasure to pay for a marginal clocker at inflated 'scarcity' prices... When the time comes, I hope Nvidia has released its 7nm consumer versions (ie 3080 Ti, 3090, even Titan ?) and AMD's 6900 XT will be out there aplenty on the shelves

Hi 'E; - pls check pm

banned coz...success at trying to forget = don't remember

The One Plus Nord isn't even available over here yet (only beta sign-up, per below). Since I don't like Apple ('it's [marketing bumf] magic' ?), I'd look at a mid-range Samsung Galaxy 5G, but I'm anything but an expert on mobile phones. ...also, if you like things called 'Nord' with great specs for the money, how about this 142 meter mega yacht; just finishing sea trials...it has all kinds of mobile communication built in at no extra charge !

...amnesia can be a beautiful thing ? ban

banned for not driving electric...self-banned for not driving electric

...yeah, I know those frustrations (I built at least five past quad-SLI / quad-fire system and lost count of dual and triple GPU setups). But this is about post-SLI/Crossfire (or what is traditionally known as SLI, such as AFR/2) and looking at upcoming mGPU technologies to 'replace' such traditional SLI with tech that has multiple GPU 'chiplets' or 'tiles' on one card that can 'appear as' a single GPU For now, I am having a grand old time via 2x 2080 Ti in CFR in Microsoft Flight Simulator 2020 (EHW thread on that > here ) - got CFR working with that shortly after Flight Sim 2020 release...4K Ultra ?

...it's the first commercial release (rather than rumour), albeit for the lucrative enterprise market...specifically for streaming. Each card has 4 Intel Xe Lp GPU 'tiles' but with 8 GB VRAM each (32 GB total) and an interesting interconnect*. Apparently, 4-way card configurations (= 16 Xe Lp) are possible. Ecosystem and driver development (including for Linux) seem thorough. This is not your home gamer competitor to 6900 Xt and/or 3090 yet (which is currently known as Intel Xe HPG), nevertheless a step forward for mGPU, IMO. There is talk of an Intel discrete 4-tile 'gamer' card competitor (Xe HPG) in the works, but when / how much, who knows. Still, the first commercial mGPU products are hitting the enterprise market, and that seems like a necessary step on the way for HPG consumer variants, not least as it seems to suggest that the Embedded Multi-die Interconnect Bridge (EMIB) high density interconnect is working - the tech that makes it appear as a single GPU and avoid traditional SLI / Crossfire issues.

A quick update / November 2020...'for your viewing pleasure', the Microsoft Flight Simulator 2020 thread focuses on NVlink / SLI / CFR results > here Also, per earlier posts in this thread, some stirrings in the multi GPU / single card field...Intel Xe / server apps (for now)...

...I used to use that foam insert as a 'test-bench' when working with DICE or even LN2...just cut out an opening at the back of the mobo where the CPU sits. That usually worked fine, unless you had a really heavy (or multiple) pots that pressed some of the caps on the back of the mobo into the foam if they mobo didn't have a shield. Foam removal-from-mobo technique was critical ?

...whether they lean on their own fabs or go with TSMC per below, Intel had made a set of major misstep several years back (under the previous CEO) which has changed their corporate trajectory down. In addition to CPU, a lot will also ride on the Xe tile graphics for the enterprise market. In any case, I certainly believe that Intel will eventually recover, but re. current and next step fab tech execution, TSMC leads by a country mile (and not just Intel). Now, as the previous 7 or so years had shown, it is best not to have just one price setting and product cycle setting, dominant player - from that consumer perspective alone, I hope Intel will regain its strength. Trouble is, there's a long lead-time in that business, and mistakes made 4+ years ago have an impact for a long while after.