PTT評價

[情報] AMD Radeon RX 6800, 6800 XT, and 6900

看板PC_Shopping標題[情報] AMD Radeon RX 6800, 6800 XT, and 6900 作者
chirex
(請叫我大魔道士)
時間推噓50 推:50 噓:0 →:73

https://reurl.cc/Md9br3
AMD Radeon RX 6800, 6800 XT, and 6900 XT aim at Nvidia's best—even the RTX
3090
Hello, RDNA 2. Here's everything AMD revealed about the Radeon RX 6000-seriesgraphics card, and the new tech inside them.

By Brad Chacos
Senior Editor, PCWorld | OCT 28, 2020 9:28 AM PDT

Welcome back to the high end, AMD.

For the first time since the days of the Radeon R9 290X—seven long years ago
—AMD’s Radeon group is prepared to unleash enthusiast-level graphics cards
that can do battle with Nvidia’s best and brightest. During the company’s
second “Where Gaming Begins” event on Wednesday, AMD revealed a trio of
Radeon GPUs based on a new RDNA 2 architecture, all armed with “
revolutionary Infinity Cache” technology, sky-high clock speeds, and
interesting synergies with Ryzen. They take direct aim at Nvidia’s RTX
30-series offerings, culminating in a $999 Radeon RX 6900 XT that seeks to
seize the gaming crown from the monstrous $1500 GeForce RTX 3090 when it
launches December 8—for $500 less than Nvidia’s beastly GPU costs.

The $579 Radeon RX 6800, meanwhile, will go toe-to-toe with the $500 RTX 3070releasing tomorrow, and the $649 Radeon RX 6800 XT intends to challenge the
$700 RTX 3080. Both launch November 18, about two weeks after AMD’s Ryzen
5000 CPUs hit the streets.

[ Further reading: Best graphics cards for PC gaming ]
ultimate gaming platform
AMD
While Nvidia’s recent gaming GPU releases went light with raw memory
capacity, opting for an upgrade to faster GDDR6X VRAM instead, AMD equipped
every high-end Radeon RX 6000 card with a substantial 16GB of standard GDDR6,plenty to handle 4K gaming well into the future. And while Nvidia’s RTX 3080
and 3090 consume much more power than previous GeForce incarnations, AMD
claims it’s achieved its lofty goal of a 50 percent performance-per-watt
increase compared to the original RDNA architecture, which makes the Radeon
RX 6000-series XT graphics cards less power-hungry than their Nvidia rivals.

Yes, you read that right: Radeon GPUs will need less power than competing
GeForce cards. My, how the tables have turned.

We’ll start by diving into the raw specs and performance claims for the
Radeon RX 6800, 6800 XT, and 6900 XT, because we know that’s what you’re
here for, but stick around afterward when we dig into key changes in the RDNA2 architecture and new Radeon ecosystem features, like Smart Access Memory
that ties into Ryzen 5000 CPUs and AMD’s new Infinity Cache. Those provide
crucial additional context for some of the slides you’re about to see.

TABLE OF CONTENTS
Radeon RX 6800 vs. GeForce RTX 3070
Radeon RX 6800 XT vs. GeForce RTX 3080
Radeon RX 6900 XT vs. GeForce RTX 3090
RDNA 2 and Infinity Cache
Ryzen synergy, Rage, and DirectX 12 Ultimate
Radeon RX 6800 vs. GeForce RTX 3070
radeon rx 6800
AMD
Let’s start at the more affordable end.

The Radeon RX 6800 packs 60 compute units and a 1,815MHz game clock—90MHz
higher than the rival GeForce RTX 3070’s rated Boost speed. More
importantly, AMD’s card triumphs in memory capacity, packing a full 16GB of
GDDR6 RAM versus the 3070’s 8GB. Nvidia pitched the 3070 as the ultimate
1440p graphics card, but it can do well in 4K games, too. That 8GB memory
buffer probably won’t hold up well for 4K gaming as the years go on though,
as some games already exceed it. The Radeon RX 6800 won’t suffer the same
problem. Nvidia’s card enjoys a slightly lower total board power rating.

radeon rx 6800 4k
AMD
radeon rx 6800 1440
AMD
Now things are getting interesting. Vendor benchmarks should always be taken
with a grain of salt, and you should always wait for independent reviews of
pricey hardware, but AMD’s data tends to be reliable—and it shows the
Radeon RX 6800 beating the GeForce RTX 3070 across the board here, at both 4Kand 1440p resolution. Wow. Radeon chief Scott Herkelman told me that AMD’s
lab used identical systems for all tests, changing only the graphics card,
and that the company used the best-performing API for a given graphics card
in all tests. So if Nvidia’s GPU performed better in a game using DirectX
11, and AMD’s exceled with DX12, these results reflect apex performance.

It’s not quite as cut-and-dry as it looks, however. See that tiny “+Smart
Access Memory” next to the Radeon RX 6800’s name at the bottom? This graphs
shows AMD performance with that feature active, which can provide an
additional performance boost when you drop a Radeon 6000 GPU into a system
with a Ryzen 5000 CPU installed. We’ll get into it more later, but these aren’t stock results for the RX 6800.

Radeon RX 6800 XT vs. GeForce RTX 3080
radeon rx 6800 xt
AMD
Stepping up the stack, the Radeon RX 6800 XT increases the compute unit countto 72, game clock speeds to a hair over 2GHz, and the total board power to
300 watts. It also offers a roomy 16GB of memory. By comparison, Nvidia’s
GeForce RTX 3080 has a Boost clock nearly 300MHz slower (though it tends to
run faster in reality) and just 10GB of memory, though it uses the faster
GDDR6X variety for more bandwidth. The Radeon RX 6800 XT’s 300W power rating
is 20W less than the RTX 3080’s.

radeon rx 6800 xt 4k
AMD
radeon rx 6800 xt 1440
AMD
In these results, AMD isn’t putting its thumb on the scale with Smart Access
Memory. In its stock state, the Radeon RX 6800 XT meets or flat-out beats theGeForce RTX 3080 in 7 of the 10 benchmarks here at 4K resolution, where Nvidia’s card performs its best. Drop the resolution to 1440p, where Nvidia’s
Ampere can’t take full advantage of its abundant shaders, and AMD opens its
lead even more, triumphing in every game except Division 2 and Resident Evil
3. Hot damn.

Radeon RX 6900 XT vs. GeForce RTX 3090
radeon rx 6900 xt
AMD
Now we get to the battle of the (almost) Titans. The Radeon RX 6900 XT packs
identical specifications to the Radeon RX 6800 XT except one, boosting the
critical compute unit count from 72 to 80. Yes, it still runs over 2GHz, and
yes, it still sticks to 300W of total board power—50W less than the
power-sucking GeForce RTX 3090.

Nvidia’s marketing for the 3090 revolved around 8K gaming and the card’s
massive, creator-friendly 24GB of GDDR6X memory. The company even calls the
step-down RTX 3080 its gaming flagship—not the 3090. AMD is keeping the
focus on normal gamers with the Radeon RX 6900 XT. This card comes with 16GB
of standard GDDR6 memory, just like the other Radeon GPUs revealed today. It’s not as spacious and not as fast as the RTX 3090’s memory, but 16GB of
GDDR6 should hold up well for 4K gaming over the long term nonetheless.
Opting for a less extreme VRAM setup means AMD can keep the cost of the
Radeon RX 6900 XT much lower compared to its rival.

radeon rx 6900 xt 4k
AMD
So many people thought this was impossible, but here we are—at least in AMD’
s supplied benchmarks. The Radeon RX 6900 XT beats or ties the GeForce RTX
3090 more often than not at 4K, and substantially so in Forza, with all gamesblowing well past the golden 60-frames-per-second mark even with their
highest visual settings enabled.

Note that AMD is tipping the scales again here, and more so than before. Not
only is Smart Access Memory active on the Radeon system, but so is Rage, AMD’s new automatic overclocking tool in Radeon Software. A victory is a victory,but you can’t help but wonder how much extra performance those features are
adding to the top. A slide you’ll see later shows the duo improving
performance by anywhere from 2 to 13 percent depending on the game, with an
average uplift of 6.4 percent across eight games.

It’s also worth noting, however, just how much cheaper the Radeon RX 6900 XT
is compared to the luxurious GeForce RTX 3090. $500 ain’t chump change.

That’s it as far as the Radeon RX 6000-series graphics cards themselves go.
For a deeper look at the RDNA 2 architecture and new features inside AMD’s
high-end GPUs—including Rage, Smart Access Memory, and Infinity Cache—keep
reading.

RDNA 2 and Infinity Cache
per per watt
AMD
AMD knew that to catch up to Nvidia, its GPUs needed to become much more
efficient than they used to be. RDNA 1 achieved a 50-percent power
improvement over previous Radeon GPUs, but a large part of the efficiency
gains came from moving to the advanced 7nm manufacturing process. The RDNA 2
architecture stays on 7nm, so AMD’s engineers needed to boost its efficiency
by another 50 percent in other ways.

compute units
AMD
It started with the compute units—the beating heart of the GPU. AMD’s
graphics chief engineer, Laura Smith, said the company went through the
design with a fine-tooth comb to root out inefficiencies: rejiggering data
paths, rebalancing pipelines, and (in a page from the Ryzen design handbook)
enabling “pervasive fine-grain clock gating,” which is an interesting twist
I’m looking forward to hearing more about, as GPUs behave very differently
than CPUs. The new compute units are 30 percent more efficient than before,
which Smith says is crucial for scaling RDNA up to these enthusiast-class
GPUs.

frequency
AMD
AMD’s engineers also managed to squeeze much higher clock speeds out of RDNA
2 on the performance side of things. Where the Radeon RX 5700 XT sported a
1,755MHz game clock, AMD says the Radeon RX 6800XT and 6900XT should sustain
2,015MHz speeds during gaming, and can even go to 2,250MHz in other tasks.
(Smith actually says the Radeon RX 6900 XT’s power-per-watt increase versus
RDNA 1 was even higher, at 65 percent.)

infinity cache
AMD
But the most intriguing—and mysterious—power gains come from the new 128MB
“Infinity Cache” built into the Radeon RX 6000-series. Smith says the
Infinity Cache is a high-density design that has its roots in “Zen 3” Ryzen
’s cache decisions, but rearchitected for Radeon’s gaming needs. Details of
what Infinity Cache actually does remain murky, but Smith says it basically
allows Radeon RX 6000 graphics cards to deliver twice the memory bandwidth atlower power cost. Including Infinity Cache lets AMD stick to a smaller
256-bit bus for the Radeon RX 6000-series, which provides substantial power
savings over moving to the larger bus sizes you normally see in high-end
gaming cards. We can’t wait to hear—and test—more.

Ryzen synergy, Rage, and DirectX 12 Ultimate
Hardware can’t run without software, and part of AMD’s presentation was
dedicated to showcasing some nifty new capabilities found in Radeon RX 6000
graphics cards. Most intriguing? AMD Smart Access Memory.

amd smart access memory
AMD
Like AMD’s Smart Shift technology in laptops (and the RDNA 2-based
PlayStation 5), AMD Smart Access Memory is designed to give you better
performance if you’re all-in on AMD’s hardware. If you have both a Ryzen
5000 CPU and a Radeon 6000 GPU in your system, AMD Smart Access Memory gives
your CPU access to your graphics card’s full memory buffer, rather than the
usual 256MB chunks. That direct communication cuts down on buffering and
potential latency. Even though game makers haven’t had a chance to optimize
for the closely guarded tech yet, Radeon chief Scott Herkelman told PCWorld
that simply flipping this feature and Rage overclocking on simultaneously in
AMD’s labs resulted in performance increases across a series of games, with
Forza Horizon 4 running 13 percent faster.

amd smart access memory and rage
AMD
Fascinating stuff. As with Infinity Cache, we’re looking forward to seeing
deeper details, but several AMD-supplied benchmarks for the Radeon RX 6000
GPUs show results from numerous games with Smart Access Memory active.

AMD’s new cards will also support Microsoft’s upcoming DirectStorage API,
which will let your GPU talk directly to your NVMe SSD for vastly improved
loading times in games. Getting the various parts of your PC to talk directlyrather than flowing through system RAM suddenly seems like the hot new trend
for 2021. (Nvidia’s RTX 30-series also supports DirectStorage, using “RTX IO
” branding.)

rage
AMD
AMD is also spiffing up the one-click overclocking capabilities found in its
Radeon Software. Going forward, you’ll find a new “Rage” preset tuning
option that applies an automatic overclock tuned to your specific card—a
helpful tool indeed for people who just want their games to play as quickly
as possible with no headaches. Herkelman told me there are some slight
differences compared to the previous automatic overclocking feature offered
by Radeon software, but he didn’t get into specifics.

Fun fact: The “Rage” name is a shoutout to a popular line of graphics cards
released by ATI in the early days of 3D acceleration, before AMD purchased
the company. ATI’s “Fury” moniker made a comeback in graphics card form a
few years back, and now Rage will be back on every Radeon-powered PC. I dig
it.

radeon boost
AMD
Speaking of Radeon Software features, AMD also briefly teased how its
existing “Radeon Boost” and “Radeon Anti-Lag” features combine to improve
latency in games that demand quick responses, especially when paired with a
FreeSync monitor. GeForce cards just gained the Nvidia Reflex suite of
latency boosting features, so it’s clear that game responsiveness is
becoming a new battleground for graphics card makers.

dx12 rdna2
AMD
Finally, like the GeForce RTX 30-series, AMD’s Radeon 6000 graphics cards
will be fully DirectX 12 Ultimate-compliant. Microsoft calls DX12 “a force
multiplier for the entire gaming ecosystem” by unifying an array of new
features—mostly ones introduced in Nvidia’s RTX 20-series, but largely
ignored by developers—across all PC and next-gen Xbox hardware. “When Xbox
Series X releases, there will already be many millions of DX12 Ultimate PC
graphics cards in the world with the same feature set, catalyzing a rapid
adoption of new features, and when Xbox Series X brings a wave of new consolegamers, PC will likewise benefit from this vast surge of new DX12 Ultimate
capable hardware,” Microsoft said when announcing the API.

Being DirectX 12 Ultimate-compliant means the Radeon RX 6000 graphics cards
will support variable rate shading, mesh shading, sampler feedback, and yes,
real-time ray tracing—though AMD didn’t go into details about its hotly
anticipated ray tracing implementation.

dx12u fidelityfx
AMD
AMD wasn’t entirely mum, though. Nvidia has a clear lead on ray tracing, and
the company’s proprietary Deep Learning Super Sampling (DLSS) plays a big
role. The AI upscaling feature leverages the dedicated tensor cores in
GeForce GPUs to speed up frame rates, counterbalancing ray tracing’s
performance impact—but remember, it’s an Nvidia-proprietary technology. AMD
teased a solution of its own as part of the company’s open FidelityFX tools
during today’s presentation. Tucked away in the corner of the slide, there
was a small box that simply said “Super Resolution.”

Herkelman wouldn’t go into details, but he confirmed to PCWorld that this
will be AMD’s open response to Nvidia’s DLSS, designed to work on any
graphics hardware—even APUs—and across various platforms, because AMD’s
RDNA 2 GPUs also power the next-gen PlayStation 5 and Xbox Series X consoles.

That broad adoption of AMD technologies could be an ace in the hole for Radeon’s ray tracing future. The new consoles will be getting off the ground at
roughly the same time as the Radeon RX 6000 series, all bearing RDNA 2 GPUs.
With the Radeon RX 6000 series, Xbox Series X, and even Nvidia’s RTX
30-series all driving DirectX 12 Ultimate forward in the same time frame as
well, AMD is betting that DX12U’s features will become ubiquitous. Everyone
is working on the same page now.

amd consoles
AMD
RDNA 2 isn’t just for the Radeon RX 6000 graphics cards.
As the graphics chip inside the consoles, AMD’s yet-to-be-revealed ray
tracing implementation could become the standard target for game developers
despite Nvidia’s two-year lead and impressive proprietary hardware. If AMD’
s ray tracing becomes the standard, its open-source “Super Resolution”
technology inside of FidelityFX could become the standard as well. DLSS 2.0
is black magic, but it’s locked to Nvidia GPUs on PCs, and requires custom
code.

Of course, as with Infinity Cache and AMD Smart Access Memory, it depends on
the details. AMD isn’t revealing anything more about Super Resolution today.

Bottom line, though? After seven long years, AMD’s RDNA 2-equipped Radeon RX
6000 graphics cards are finally ready to bring the fight to Nvidia at the
very edge of enthusiast-class performance. We’ll need to see if AMD’s lofty
claims hold up to scrutiny by independent testers, but we can’t wait to get
our hands on these graphics cards. Competition is a wonderful thing—and long
overdue at the high end. Welcome back, AMD.

Note: When you purchase something after clicking links in our articles, we
may earn a small commission. Read our affiliate link policy for more details.Related:Graphics Cards Gaming AMD PCs Hardware.


===
直播詳細記錄出來了。

簡單就是

RX 6800 的對手是 RTX 3070

RX 6800XT 的對手是 RTX 3080

RX 6900XT 的對手是 RTX 3090

RX6000系列 VRAM 全上16GB…

7nm製程,台積製造,VRAM 16GB...有夠香,

感謝蘇媽讓我們可以買到便宜的Nvidia(咦?)

--

想想…,蘇媽這算軟體(光追)不打算跟老黃拼,

直接用硬體輾嗎?
我還沒說完吶~~ 夠多了夠多了,回家吃藥了啦~!
    \ /
    ○
    ︵ \○╱/|
    . . ﹎ ﹍﹍ ╱ ̄﹚╱>


--

※ PTT 留言評論
※ 發信站: 批踢踢實業坊(ptt.cc), 來自: 114.46.41.200 (臺灣)
PTT 網址

lope304 10/29 01:25我不想努力了

qweertyui89110/29 01:268G10G還敢丟出來賣啊老黃,現在要怎麼切

Shigeru777 10/29 01:26我要賣掉3080了 香

rtwodtwo 10/29 01:26買不起.... 有沒有300鎂等級的

aaron5555 10/29 01:27怎麼切? 價格往下砍就好啦 反正amd用GG根本沒辦法

aaron5555 10/29 01:27打價格戰

kimisawa 10/29 01:288G不就跟intel 的祖傳4C8T一樣嗎 XD

LinYiJen 10/29 01:28坐等EVGA 3080補貨我可以再買一張

kimisawa 10/29 01:28你覺得老黃可以打價格嗎? 3070根本就是3080/3090打

LinYiJen 10/29 01:29這次應該買得到了吧

kimisawa 10/29 01:29下來的,所以說老黃高階價格也壓不下來還有GDDR6X

Marlboro27 10/29 01:29中國價格 6800/4599 6800XT/5099 6900XT/7999

sigma9988 10/29 01:29降價再出大vram suuuuuper版

paio 10/29 01:30那不就要18000-20000自製卡 ROG 也要兩萬了嗎

shepherd71 10/29 01:30明天3070的售價有機會改嗎xdd

reticentcrow10/29 01:31老黃的刀法失效啦 以前是性能追不上才有得切

paio 10/29 01:31EVGA同樂會不是現場都會便宜1800嗎上面說的

tonyii 10/29 01:313070不買了www 說扳倒2080ti 這老黃做生意不老實阿

paio 10/29 01:32評測說互相上下2080Ti但是6800說有扳倒2080Ti

flipflap 10/29 01:33A粉開始胡說八道了 3080/3070的die size就不一樣 30

flipflap 10/29 01:3380是要怎麼打下去變3070 用縮小燈嗎

a3831038 10/29 01:34沒,看評測3070幾乎都是小輸2080ti,只有某些專業軟

a3831038 10/29 01:34體會贏

friedpig 10/29 01:343080要變3070又沒啥問題 老黃賠大本而已 又不是沒幹

friedpig 10/29 01:34

※ 編輯: chirex (114.46.41.200 臺灣), 10/29/2020 01:36:37

Shigeru777 10/29 01:34反正老黃現在問題是沒貨 再不逼出產能我看之後也不

Shigeru777 10/29 01:35用賣了

tonyii 10/29 01:352080ti升級驅動在路上了 優化完3070就贏2080ti了(誤

otaku690 10/29 01:35有沒有HDMI 2.1啊

AZBTPATONY 10/29 01:36搶3080的黃牛直接爆死

friedpig 10/29 01:36不過大家怎麼都覺得AMD就一定產能充足了(?

aaron5555 10/29 01:363070又不是用貴貴的gddr6x 價格有什麼不能砍的 根

aaron5555 10/29 01:36本只是想訂多少跟賺多少而已

friedpig 10/29 01:37老黃很少直接砍價格的拉 很丟臉變相承認輸了

pipi5867 10/29 01:37老黃太有本錢打價格戰了..

friedpig 10/29 01:37通常是出個S或TI加點規格變相砍價

knml 10/29 01:38老黃都是用切的,切出一個跟你玩

kashima228 10/29 01:38說AMD產能不足是看不起台GG星人嗎?

aaron5555 10/29 01:38欸 說到產能 那個4000系apu到底要不要解封組裝限定

aaron5555 10/29 01:38啊 不也是同樣用gg7nm?

marunaru 10/29 01:39GG:一個月加班150小時產能不足那就加200小時

friedpig 10/29 01:39接下來應該馬上出3070TI(S) 600鎂 3080TI(S)1000鎂

friedpig 10/29 01:39來扛了

marunaru 10/29 01:394000是下單的量本來就不多

Nexus5X 10/29 01:39中科的台G仔都加班到躲過氣爆了 現在應該都睡公司

friedpig 10/29 01:39GG產能有更重要的Console在搶 難講喔

a3831038 10/29 01:39幫朋友拿了顆4650g,真的蠻好用的

friedpig 10/29 01:40AMD顯卡這邊他們自己量還是不好估 打平盤根本不確定

friedpig 10/29 01:40能賣多少 下的量未必會多敢衝 4000就很保守過了

pipi5867 10/29 01:41產能真的很難講..以往新顯卡或新cpu都很容易缺貨

aaron5555 10/29 01:41不是啊 既然下單的量不多那應該可以解封了吧...

pipi5867 10/29 01:41別忘了以前也一堆都GG代工的

sigma9988 10/29 01:43gg已招募更多新鮮的肝了但還不夠

benqlove 10/29 01:456800全部打贏2080ti顯存又多,貴個79差不多

AreLies 10/29 01:46顯存

AreLies 10/29 01:46https://i.imgur.com/rZjCQM9.png

a3831038 10/29 01:48紅綠燈是三小

ltytw 10/29 01:48老黃現在又來不及排gg生產gg版的ga晶片,有大半年

ltytw 10/29 01:48時間只能打價格戰

pipi5867 10/29 01:50支語警察吧xd

HotDogCC 10/29 01:56沒差拉 反正下一代顯卡出之前光追遊戲我看大概也不

HotDogCC 10/29 01:56會多到哪裡去

arrenwu 10/29 01:57痾 妳貼這一大篇都是英文的 我看不懂怎半?

chirex 10/29 01:58點進去連結看圖片就好

gibala 10/29 02:02老黃:來不及切訂單都取消了

aaron5555 10/29 02:02這網站廣告也太多 一點進去就送我一個全螢幕廣告+

aaron5555 10/29 02:02網站中間廣告+實況內容 煩死= =

chirex 10/29 02:03有廣告?我用PC開有裝AdGuard,所以不知道有廣告

aaron5555 10/29 02:04我用手機點的 廣告多到我手機在lag...

chirex 10/29 02:04是喔?那真抱歉,我沒注意到這點

aaron5555 10/29 02:05沒事 是網站的問題不是你的問題

e2167471 10/29 02:0916GB 是不是該把30卡的價差拿去買AMD+144比較正確

Marlboro27 10/29 02:12話說xv272up 9K找一塊 特價中

e2167471 10/29 02:14這...? 小卡的靈壓...!

e2167471 10/29 02:18查了一下這咖好像沒有不閃屏?

helba 10/29 02:219千1萬塊級只有兩個選擇 xv272up vg27aql1a

helba 10/29 02:22群創K7b面板

ctes940008 10/29 02:34建議你把中文結論放在最前面,手機板閱讀很痛苦。

cs22420628 10/29 02:45蘇阿姨我不想努力了

nicholassys 10/29 03:12沒有DLSS怎麼和nvidia打…FPS差那麼多

Banrin 10/29 03:18重點是價格,產能不足對手切不了,希望可以打價格戰

atbb 10/29 03:26看來各大通路的30x 明天開始都會有貨了

jior 10/29 03:58聽說這次頻率還能上去,讓人期待非公板

Gan0liang 10/29 04:36DLSS是給4K以上用的 大部分人根本無感

aegis43210 10/29 06:18蘇媽給的價格不漂亮呀?不值得開心

kimula01 10/29 06:24台灣到底哪時候上市啊

torpp 10/29 07:59你無感不代表大部分人都無感

pxhome 10/29 08:04NV會改價格嗎? 記憶體容量大輸

sina1 10/29 08:05幸好老黄這次開價新卡沒有比前代高,不然剛出就跳水

SaberMyWifi 10/29 08:298G跟16G比 我小班姪子閉著眼睛都知道怎麼=.=

SaberMyWifi 10/29 08:29知道怎麼選

kuma660224 10/29 08:52A早就有CAS能取代DLSS了 用不同原理

vocaloid02 10/29 08:53紅綠燈....笑死XD

jerrychuang 10/29 08:59AMD CPU都加價了,顯卡更有底氣打價格戰了

kuma660224 10/29 09:01其實7nm算毎億萬電晶體成本已經不貴

kuma660224 10/29 09:03Navi21電晶體甚至略少於GA102

kuma660224 10/29 09:04用GDDR6也比GDDR6X便宜

kuma660224 10/29 09:06理論上蘇媽完全有能力跟老黃拼價格戰

kuma660224 10/29 09:07互相傷害.....所以把3090拉下來打

kuma660224 10/29 09:09蘇媽唯一吃虧的是要等家機備料到夠

kuma660224 10/29 09:09所以RDNA新卡晚了幾個月上

kuma660224 10/29 09:10但GG實在太神 旗艦大Die輕鬆跑2ghz+

kuma660224 10/29 09:11電晶體預算小輸的 居然能下剋上

kuma660224 10/29 09:13老黃選三星+GDDR6X 就變成不利之舉

kuma660224 10/29 09:13GG: 我最討厭想壓價不忠誠的客戶.

kuma660224 10/29 09:14多幫AMD Tuned了幾下

buteo 10/29 09:19蘋果和AMD忠誠度和量都高 應該會好好服務

tsairay 10/29 09:49AMD接下來只要供貨充足,老黃就危險啦

AKSN74 10/29 10:41等等,那這樣***到底有沒有產能問題?

AKSN74 10/29 10:41這幾波新聞與消息看下來搞得有點亂啊

flipflap 10/29 11:03看到有人說蘋果忠誠度高?我笑了

kuma660224 10/29 11:20蘋果是吃過虧 才忍住不玩雙商源

friedpig 10/29 11:20說到Apple最近蠻多討論N5高頻烙賽的問題

kashima228 10/29 11:45A14那個是蘋果的問題吧

friedpig 10/29 12:06麒麟也不太好拉

ray2000 10/29 12:11下代又提升50%的話 要別人怎麼活

MrDisgrace 10/29 12:42多幫AMD Tuned了幾下 wwwwwwwwwwwwwww

ms0545173 10/29 13:5616GB真的香

kuma660224 10/29 14:27蘋果手機高溫降頻一直是常態吧

kuma660224 10/29 14:27它毎代都不太在乎散熱導熱

kuma660224 10/29 14:28它SOC要放到散熱空間大的ipad才出全力

israelii 10/29 15:27講中文

axakira 10/31 21:26哈哈哈,一堆臉被打腫的盤子和黃牛