MSI GeForce RTX 3080 Gaming X Trio Evaluation

Please log in or register to like posts.

Producer: MSI

UK place (as reviewed): MSRP £759.99 (inc. VAT)

US place (as reviewed): MSRP $749.99 (exc. tax)

Having now not been sampled by Nvidia for the RTX 3080 originate, we’ve now not been ready to conceal it the come we would beget cherished. Lastly, though, we’re kicking things off with a accomplice card overview courtesy of MSI, which has sampled us the RTX 3080 Gaming X Trio. Offered that it’s our first RTX 3080 overview, we’ll evaluate it to the RTX 2080 (which launched at the identical MSRP of $700) to evaluate the gen-on-gen improvement; the flagship card of the favored 10 Series, the GTX 1080 Ti, to abet Pascal customers work out if now will most likely be the time to upgrade; and the RX 5700 XT, AMD’s closest offering at most modern, to survey staunch what roughly hole AMD now has to shut with the upcoming RX 6000 Series.

As MSI’s flagship RTX 3080 (for now), the Gaming X Trio comes in above MSRP, particularly £759.99 in the UK, so we’ll be diving into the components and the thermals to enable you to evaluate whether or now not this £110 / 17% top class will most likely be justifiable. Sooner than that, though, let’s give the card and the recent GPU some context.

Since right here’s our first overview coverage for any Ampere GPUs, we’ll rapid sweep thru the major architectural and hardware adjustments. Offered that Ampere is an evolution of Turing, a background realizing of that structure would be invaluable – thankfully now we beget a detailed article to web you in regulate.

  • Contemporary task: Nvidia is now using a custom model of Samsung’s 8nm task to diagram Ampere GPUs. This come bigger transistor density and more vitality efficiency (performance per watt).

  • More CUDA Cores: Nvidia has restructured the Streaming Multiprocessors (SMs), maintaining the devoted FP32 (CUDA) datapath however turning the devoted integer datapath from Turing into one which might bustle both integer OR FP32 calculations (now not both at this time). Attainable FP32 processing vitality is as a result of this fact twice what it became (per SM), however the profit of this in games (where FP32 and integer calculations are wanted concurrently) will rely on the workload.

  • Contemporary RT and Tensor Cores: The RT and Tensor Cores are now the 2nd Gen and 3rd Gen versions, respectively. On the RT facet, Nvidia has doubled the triangle intersection hardware to pork up performance in that metric and added a unit to abet in scenes with both ray tracing and circulate blur. The recent Tensor Cores, meanwhile, are sooner by 2x for processing dense matrices and by 4x with sparse matrices. With Ampere having half as many Tensor Cores per SM, each and each SM is thus as a minimal as like a flash as earlier than and up to 2x the bustle with more favourable workloads.

  • GDDR6X: The RTX 3090 and 3080 both characteristic GDDR6X memory, which is itself the final end result of a partnership between Nvidia and Micron. Performance and efficiency enhancements in the accomplish bewitch us to even bigger heights in phrases of both the memory recordsdata rate (GDDR6X is specced to bustle at up to 21Gbps!) and the GPUs’ total memory bandwidth – a must beget for “feeding the cores”.

  • PCIe 4.0: Following AMD, Nvidia’s most modern batch of playing cards beget now migrated to the PCIe 4.0 identical old, which affords double the bandwidth of PCIe 3.0. Lawful now, sport workloads are certainly now not bottlenecked at all by PCIe 3.0 bandwidth, so don’t in actuality feel you’ll need an AMD X570 or B550 motherboard to “unlock” Ampere’s factual performance.

Whereas that’s by no come a total picture of all that’s changed, it does conceal the major adjustments that will affect the performance and advertising and marketing and marketing of the Ampere parts. It also provides you a bit context for the specs of the recent playing cards, which we’ve laid out right here:

RTX 3090RTX 3080RTX 3070RTX 2080 TiRTX 2080 AstronomicalRTX 2080RTX 2070 Astronomical
Open dateSep-20Sep-20Oct-20Sep-18Jul-19Sep-18Jul-19
Task8nm8nm8nm12nm FFN12nm FFN12nm FFN12nm FFN
Transistors28.3 billion28.3 billion17.4 billion18.6 billion13.6 billion13.6 billion13.6 billion
Die Dimension628.4mm2628.4mm2392.5mm2754mm2545mm2545mm2545mm2
CUDA Cores10,4968,7045,8884,3523,0722,9442,560
Tensor Cores328272184544384368320
RT Cores82684668484640
Texture Objects328272184272192184160
Heinous Clock1,400MHz1,440MHz1,500MHz1,350MHz1,650MHz1,515MHz1,605MHz
Boost Clock1,695MHz1,710MHz1,725MHz1,545MHz1,815MHz1,710MHz1,770MHz
Height TFLOPS35.629.820.313.411.2109.1
Reminiscence Recordsdata Rate19.5Gbps19Gbps14Gbps14Gbps15.5Gbps14Gbps14Gbps
Reminiscence Interface384-bit320-bit256-bit352-bit256-bit256-bit256-bit
Reminiscence Bandwidth936GB/s760GB/s448GB/s616GB/s496.1GB/s448GB/s448GB/s

Despite the architectural adjustments, the RTX 3000 Series is unexcited in half one thing of a brute-power come to raising performance, and the RTX 3080 is a ideally suited example. Whereas its GA102 GPU is nowhere shut to completely-enabled (now not even RTX 3090 has this), it is a long way a noteworthy bigger and denser GPU than the RTX 2080’s TU104, and boy, does it showcase. There are spacious will enhance to the counts of CUDA Cores, Texture Objects, and L2 cache, all of which would be invaluable in any sport, and the elevated ROPs and the usage of more and sooner memory with two extra memory controllers come the support-cease has been suitably widened as properly. Comparing numbers love this when the structure has changed isn’t especially invaluable for indicating performance variations, useless to claim, however it certainly serves to illustrate the ‘more of the entirety’ come that Nvidia has taken.

For RTX-particular workloads, there’s a 48% amplify in RT Cores (which is ready to bustle sooner too, be conscious), and whereas Tensor Cores are clearly the part that needed to give, having diminished from 368 to 272, make a selection that every and each is 2-4x sooner than earlier than – performance is unexcited up.

The final half of the performance puzzle is clock bustle, and the RTX 3080 is a match for the RTX 2080 right here, as a minimal on boost bustle. The specs above accomplish showcase the reference-level speeds, it can well unexcited be licensed; the RTX 2080 Founders Edition had a 90MHz bigger boost clock than reference, however Nvidia isn’t doing the kind of shenanigans with this originate. The lower abominable clock suggests a less aggressive curve on the recent half, however most continuously speaking, Nvidia is working a hell of noteworthy more hardware at the identical(ish) clock speeds.

All of which brings us to vitality, which is the remark kicker: The RTX 3080 desires juice the come Jensen desires leather. The TDP at 320W is big, a total 105W (49%) bigger than the RTX 2080 and 70W more than even the RTX 2080 Ti. That’s now not to tell the recent task isn’t doing the leisure – we shudder to reveal what the identical GPU constructed on 12nm FFN would cost in vitality (and size) if it had been even doable – however rather to highlight that right here’s a ‘more with more’ originate in wish to a ‘more with less’ one.

This dramatic amplify in vitality consumption necessitated a rethink of the Founders Edition cooling resolution. The RTX 3080 FE uses a in actuality tiny PCB to leave as noteworthy room as doable for cooling, and the spacious change is useless to claim the usage of self reliant push-pull fans. With one fan pulling air thru the cooler and expelling it out the assorted facet, Nvidia’s accomplish in actuality alters case airflow dynamics.

Since we don’t beget a Founders Edition card, our recordsdata on it comes from relied on sources interior our family. The consensus so a long way appears to be like to be to be largely definite. The industrial accomplish has received favour for diagram quality and efficiency of location (i.e. almost about the total frail location is functional and helps with cooling), and it’s even handed as a first-rate improvement in cooler accomplish. The card isn’t ideally suited, tips; the 12-pin vitality connector makes an adaptor necessary and cable routing untidy (for now), and overclocking headroom appears to be like to be very minute as properly.

With Nvidia already using a formidable cooling resolution and leaving the RTX 3080 with less overclocking headroom than we’re frail to, the accomplish and advertising and marketing and marketing teams of its board companions evidently beget had their work lower out for them. Aloof, with inquire for the card sky excessive beautiful now, consumers are seemingly in any RTX 3080 they can web their fingers on, all of which brings us – in the raze – to the card in demand, the MSI RTX 3080 Gaming X Trio, which we’ve detailed over the page.

Learn More


Already reacted for this post.

Nobody liked ?