https://tinyurl.com/as6tpy37
ANANDTECH上面有前Intel AXG部門員工的精采回應 翻譯與大家分享
I just can't believe there are people who believe Raja was delivering on GPU
promise. My other surprise is Raja was let go so late. I joined Intel's AXG
group some time back, genuinely believing Intel has plans in discreet GPUs
and datacenter ones (Ponte Vecchio, which Raja advertised a lot as a
revolution).
我真的無法相信有人會認為Raja可以實現GPU的願景。我另外驚訝的是Raja居然這麼晚才
走。一段時間前我加入Intel的加速運算系統與繪圖部門(AXG，現已解散分拆)，天真地相
信Intel對獨立顯卡和運算中心GPU(Ponte Vecchio，Raja一直大力宣傳是項革命產品)有
計畫。
As soon as I joined, I realized it was mostly marketing and my assumptions
were totally wrong. The teams were incompetent, demotivated and were running
way beyond schedules published. My boss told me long back PVC is dead, has
some serious yield issues and next best hope is Rialto bridge. Both are dead
now.
我一加入AXG，就發現這些幾乎都是廣告詞，而我的假設全都是錯的。這支團隊無能、沒
有士氣，執行速度比公告的時程慢太多。我的老闆很久以前就告訴我Ponte Vecchio沒救
了，因為有一些很嚴重的良率問題，而下一個最好的希望是Rialto Bridge。現在兩者都
掛了(Intel已宣布取消Rialto Bridge)。
But catalyst for me was one of the quarterly all hands by Raja and other
guy/gals in his org. The delays were written all over, the made some bogus
paper launch of dGPU on March 31, 2022 which was Raja's quarterly OKR. And to
add to that any further line of dGPUs were not only late, but not that
competitive also I figured. And I realized one important thing: Raja is not a
technical guy, he is a marketing guy who was blabbering about TAM and what
not for dGPUs when they were horribly lagging in execution. Check out his
ridiculously bold claims on 1000x improvement in PPA by 2027 on none other
site than Anandtech.
但對我來說，離開公司的催化劑是 Raja 和他手下其他人的季度全體會議。報告上延誤無
處不在，在 2022 年 3 月 31 日做了一些虛假的獨立顯卡紙上發布，這個發布算是
Raja 的季度目標關鍵成果。此外，獨立顯卡不只是延遲，更不如我預想的有競爭力。我
意識到一件重要的事情：Raja不是一個技術人員，他是一個營銷人員，他喋喋不休地談論
整體潛在市場，而不是討論執行非常差的獨立顯卡產品。可以去Anandtech上查看他做了
到2027年要將功率、性能跟面積(Power, Performance, Area; PPA)提高1000倍的可笑大
膽聲明。
Looking at Raja's lofty OKRs, so detached from reality, I knew he is on his
way out. That it took so long is the only surprise I see. And he seems to
have fooled Gelsinger for long. In one of the visits to overseas offices
(with Raja), Pat mentioned he sees Raja as "the guy" in GPU instead of that
nVIDIA guy in leather jacket. I almost threw up.
看著 Raja 如此脫離現實的崇高季度目標關鍵成果，我知道他即將離職。讓我驚訝的是離
職這件事居然拖了這麼久。他似乎已經愚弄了CEO Pat Gelsinger很久了。在一次Pat和
Raja共同拜訪海外辦公室時，Pat提到他認為Raja是”代表GPU的那個人(“the guy” in
GPU)”，而不是那個在NVIDIA穿著皮夾克的人。我差點吐了。
Lastly, I judge people's integrity based on their association with "like"
minded people. Raja has stake and deep collaboration with one of the shadiest
and corrupt guys in that VFX graphics company and is also a board member. How
Intel management could not see this for years is a surprise to me, seriously.
最後，我相信同類相聚，能用一個人親近的對象判斷他的誠信。Raja和一個最腐敗、最不
老實的人(位處VFX圖象公司，同時也是董事會成員)有非常深的合作關係。Intel的管理階
層居然花好幾年都沒辦法看出這點，讓我非常訝異。
Thankfully, I made a quick decision to leave Intel and could not have been
happier in retrospect. Not bitching about my ex-employer but it was a real
mess. I still wish Intel all the best and believe they still have some of the
brightest people who are not being utilized properly.
還好我很快就決定離開Intel，回顧這個決定真是做得太對了。不是要對我的前東家一直
抱怨，但那真的是一團糟。我仍然祝福Intel，相信他們仍有一些最聰明的人才而之前沒
有好好運用。